Guns and Crime

Gary Kleck. 21st Century Criminology: A Reference Handbook. Editor: J Mitchell Miller. 2009. Sage Publication.

On June 26, 2008, the Supreme Court handed down one of the most intensely awaited decisions in its recent history, holding that the Second Amendment recognizes an individual right to keep and bear arms, and not merely the right of states to maintain armed militias (see Cottrol, 1994, for a good overview of the constitutional issues linked with the gun control debate). The decision, minority opinion, and supporting briefs all cited dozens of scholarly studies bearing on the links between guns and violence. This chapter summarizes that literature.

The Use of Guns in Crime

Firearms are heavily involved in crime in America, especially homicide. In 2006, approximately 11,600 homicides were committed by criminals armed with guns, claiming 68% of all homicides (U.S. Federal Bureau of Investigation, 2008a), and an additional 100,000 to 150,000 individuals are medically treated for nonfatal gunshot wounds each year (Kleck, 1997, p. 5; see also Annest, Mercy, Gibson, & Ryan, 1995). Data from the National Criminal Victimization Survey (NCVS) indicate that as many as 500,000 violent crimes were committed in the United States in 2006 by offenders armed with guns (though not all of these involved the perpetrators actually using the guns, as distinct from merely possessing them during the incident). About 26% of robberies and 7% of assaults were committed by gun-armed offenders (U.S. Bureau of Justice Statistics, 2008).

Compared with other industrialized nations, the United States has higher rates of violent crime, both fatal and nonfatal, and a higher rate of gun ownership (Kleck, 1997, p. 64). These facts have led many people to conclude that America’s high rate of gun ownership must be at least partially responsible for the nation’s high rates of violence, or at least its high homicide rate. This belief in a causal effect of gun levels on violent crime rates has in turn led many people to conclude that limiting the availability of guns would substantially reduce violent crime, especially the homicide rate.

It is not so widely known, however, that large numbers of crime victims in America also use guns in the course of crimes, in self-defense. The best available evidence, based on 16 national surveys of probability samples of the adult U.S. population, indicates that guns are used by victims in self-protection more often than crimes are committed by offenders using guns. For example, victims used guns defensively approximately 2.0 to 2.5 million times in 1993, compared with approximately 850,000 crimes in which offenders possessed guns (Kleck & Gertz, 1995). Although some scholars have speculated that surveys overestimate the frequency of defensive gun use, there is no empirical evidence to support this conclusion (Kleck & Kates, 2001, pp. 241-264).

Scale and Patterns of Gun Ownership in America

By international standards, the share of U.S. households with guns is very high. In national surveys, 40% to 50% of U.S. households report having one or more guns; the nearest known foreign competitor is Switzerland, where about one third of households have guns, mainly because of military service requirements (Killias, 1990). There were probably more than 276 million guns in private hands in the United States at the end of 2003, about 36% of them handguns. The size of the U.S. gun stock, especially the handgun stock, increased enormously from the 1960s through the 1990s, although the share of U.S. households with guns showed little change over that period (Kleck, 1997).

One obvious policy implication of this huge existing stock is that a large supply of guns would remain available, to criminals and noncriminals alike, even if all further manufacture and importation of guns were immediately halted (Kleck, 1997, Chap. 3). In contrast, only a few hundredthousand guns are used to commit violent crimes each year. Thus, the supply of guns is hundreds of times larger than the numbers needed for criminal purposes. Consequently, even very large decreases in the supply would not produce gun scarcity but instead would merely reduce the size of the surplus. On the other hand, this does not imply that gun possession cannot be reduced among criminals or other high-risk subsets of the population, because it is possible that members of these groups can be deterred by legal penalties from acquiring or possessing guns, no matter how many are circulating.

The broad patterns of gun ownership in America do not support, in any straightforward way, the general idea that higher gun ownership rates will lead to higher violence rates, because gun ownership is generally highest in those groups where violent behavior is lowest. Although both gun ownership and violence are more frequent among males and Southerners, gun ownership is also higher among whites than among African Americans, higher among middle-aged people than among young adults, higher among married than unmarried people, higher among richer people than poor, and higher in rural areas and small towns than urban areas—the opposite of the way that violent crime is distributed (Kleck, 1997, Chap. 3).

Crime-Related Motives for Owning Guns and the Effect of Gun Levels on Crime Rates

The vast majority of Americans who own handguns own them primarily for protection against crime (63% in one national survey), and about half of all gun owners, including those who own rifles or shotguns, own them primarily for protection (Cook & Ludwig, 1997, p. 38). Still other owners cite protection as one of their reasons for having guns, secondary to hunting and other motives unconnected with crime. On the basis of the stated motives of gun owners, then, ownership of firearms is a response to crime, not just a cause of it.

This in turn suggests that higher crime rates could contribute to higher rates of gun ownership, as well as the reverse. Many research studies have provided empirical evidence that higher crime rates may indeed cause higher gun ownership rates (summarized by Kovandzic, Schaffer, & Kleck, 2005). The principal significance of this possibility is that it complicates the interpretation of research that finds more crime and violence in the same places and times as more gun ownership. It raises the question “Do more guns lead to more crime, or does more crime lead to more people acquiring guns for self-protection, or both?” When there is a possibility of this sort of two-way causation, separating one effect from the other becomes very difficult, requiring the use of highly complex statistical procedures. This chapter is not the place to address such technical matters; it suffices to say that experts continue to disagree about whether anyone has solved those statistical problems.

How Do Guns Affect Crime?

Understanding the connection between guns and crime requires appreciating three fundamental facts:

  • Whereas gun ownership affects crime in various ways, crime also affects gun ownership.
  • The possession and use of guns have both violence-reducing and violence-increasing effects.
  • The kinds of effects that possession and use of guns have on crime depend on who possesses and uses them. The effects of victims using guns for self-protection are predominantly violence reducing, whereas the effects of criminals using guns for aggressive purposes are a mixture of violence-increasing and, more surprisingly, violence-reducing effects.

Because gun effects are quite different depending on what sort of person possesses the gun, the effects of offender possession/use and victim possession/use are discussed separately. Readers should, however, keep in mind that many crime victims are themselves criminals. Indeed, crime victimization is far more common among criminals than among noncriminals, and serious violent victimization is largely concentrated among criminals. For example, research has found that over 60% of homicide victims have an arrest record. Thus, serious violence is largely a criminal-on-criminal phenomenon. It therefore would be a mistake to view the offender-victim distinction as equivalent to the distinction between wicked offenders and morally pure victims. On the other hand, it would be equally erroneous to believe that in individual incidences of violence there is no real distinction between offenders and victims or that it is impossible to tell which party is the aggressor and which is the victim. The somewhat morally unsatisfactory reality is that many of the people who are, in a given violent crime, clearly the victims of violence initiated by another person have themselves committed serious crimes in the past.

One critical implication of these facts is that even criminals use guns for genuinely defensive purposes, in incidents in which they are victims as well as for offensive or aggressive purposes in incidents in which they are offenders. Although this is almost never a part of the political debate over guns, even criminal gun possession can have violence-reducing effects as well as violence-increasing effects. Defensive uses of guns by criminals are not likely to be reported to either police or to survey interviewers, but there are nevertheless strong reasons to believe that they occur frequently and that they have the same effects as defensive uses by noncriminals.

Crime-Increasing Effects of Offender Possession and Use of Guns

Incidents of violent crime can be seen as proceeding through as many as four possible stages: (1) threat, (2) attack, (3) injury, and (4) death. The more serious a violent crime, the more of these stages the incident proceeds through. To even qualify as a violent crime, an incident must involve an aggressor, at minimum, threatening another person by word or gesture. Threat may or may not be followed by attack (i.e., an attempt to physically injure the victim). This attempt may or may not be successful (i.e., result in the victim being injured). If an injury is inflicted, it may or may not result in death. Whether the aggressor possesses a gun can affect the occurrence of each of these possible events (Kleck & McElrath, 1991).


Making a threat of violence against another person typically involves contact with another person—that is, an aggressor and victim come together in the same place at the same time. The aggressor’s willingness to confront the victim may be influenced by weapon possession, because having a weapon can give the aggressor the confidence that he or she can dominate and control the encounter and avoid being hurt himself or herself. Thus, higher rates of gun possession among prospective aggressors could increase the rate of violent encounters. There is, however, no empirical evidence directly bearing on this question.


Similarly, the aggressor’s possession of a gun could embolden him or her to go beyond a mere threat and attempt to inflict injury on the victim. A gun might also make it more feasible to successfully act on this willingness to attack, because some attacks are unlikely, or impossible, to be carried out without a gun. Many have referred to the gun as an “equalizer,” usually referring to the fact that a powerful weapon can make a victim the equal of a bigger, stronger offender. The same, however, is true for aggressive actions—an aggressor may be more willing to initiate attacks against more powerful victims because the aggressor possesses a gun. Research (Kleck & McElrath, 1991) has shown that gun use by offenders is more common in violent crimes in which less powerful aggressors attacked more powerful victims; that is, offender gun use is more common when the offenders were outnumbered by the victims; more common when women attacked men than when women attacked other women; and more common when offenders outside of the physically prime years—younger than 14 or older than 40—attacked victims in their prime years. In other words, guns seem to facilitate attacks by less powerful offenders against more powerful victims. Attacks that would otherwise have been unlikely were more feasible because the prospective aggressor possessed a gun.

Likewise, effective attacks at a distance are virtually impossible without a gun. Although little serious violence is inflicted at great distances, those that are, such as sniper attacks, virtually require gun possession to commit them. Furthermore, some scholars have speculated that some would-be aggressors would not be willing to attack others if doing so required that they do something as distasteful as coming into direct physical contact with their gun. The very fact that guns facilitate attack at a distance, even if it is a matter of a few feet, may encourage attacks by aggressors who psychologically need a more “antiseptic” mode of attack.

In addition to facilitating attacks—that is, making them possible or easier to commit—possession of guns by prospective aggressors has also been claimed to trigger attacks. Discussed in the psychological literature under the rather vague term weapons effect, this hypothesis asserts that “the trigger pulls the finger”; that is, that possession of a gun can trigger or release an impulse to aggress. The theory behind this is that if a person is already angered, and in that sense ready to aggress, even the sight of a gun, or its possession, can trigger the aggressive impulse, because of the learned association between guns and aggression. The research on this hypothesis is almost equally divided between studies supporting it and those failing to do so. The more realistic experimental studies, however, generally do not support it.

This lack of experimental support could be due to yet another effect of weapon possession, which may have the opposite effect on attack. For some people, exposure to weapons appears to inhibit aggression. If an attacker wants to injure, but not kill, a victim then possession of a deadly weapon gives him or her more ability to kill than he or she might be willing to use. Because most aggressors have less-than-lethal intentions, most of them may perceive guns this way, causing many to refrain from attacking at all rather than risk killing their victim.

Offender possession of guns also can discourage attacks by making them less necessary to the accomplishment of the aggressor’s goals. For example, a robber’s primary goal is obtaining a victim’s property, which is accomplished by intimidating the victim. Although intimidation might be achieved through an attack, it is usually achieved through threats alone—most robberies do not involve injury to the victim. One of the most strongly and consistently confirmed findings in the literature on guns and violence is that robbers with weapons are less likely to attack and injure their victims than robbers without weapons (Kleck & McElrath, 1991). By 1997 alone at least 18 studies had been conducted that, without exception, confirmed this fact. This phenomenon can be labeled a redundancy effect, because gun possession makes it unnecessary for the aggressor to actually attack the victim. Merely threatening to attack is sufficient to induce the victim to comply, because the weapon is perceived as such a lethal one. In contrast, many robbers without weapons must attack their victims at the outset of a robbery, as a way of immediately gaining control.

The redundancy effect is not limited to robbers. People committing assaults, without any intent to steal, are also less likely to actually attack their victims, instead of confining their aggression to a threat, if the assaulter possesses a gun. An assaulter’s goal may be to terrify or humiliate his victim, but if the aggressor has a gun these goals can also be achieved without actually attacking the victim. The moral irony of these facts, of course, is that guns in the hands of “bad” people have some “good” effects. This moral complexity may explain why these effects are rarely addressed in the public debate over guns. It is easier to think in black-and-white terms, and the idea that empowering bad people could have any good effects is unthinkable to some people.

Nevertheless, the evidence indicates that, when one takes into account all of these various gun effects on attack, the net effect of offender gun possession is that it reduces the likelihood of attack.


If an attack does occur, it may or may not result in injury (e.g., by a bullet reaching its target, a knife penetrating skin, or a fist or club bruising flesh or smashing bone). The attributes of weapons that can facilitate attack may also reduce the attack completion rate by encouraging attacks at a longer range, against more formidable opponents, or under more difficult conditions. It is possible to shoot a victim from a great distance, but the rate at which this is achieved is lower than the share of thrown punches that strike the victim. Regarding the more common close-range gun attacks, people unfamiliar with firearms marksmanship might assume that shooters are virtually certain to hit their target. In fact, NCVS data covering the United States from 1987 to 1992 indicate that only 18% of incidents in which an attacker shot at a victim resulted in the victim suffering a gunshot wound, whereas about 45% of knife attacks result in a knife wound (Kleck, 1997, Chap. 5). The rate of success in an aggressor inflicting injury on a victim is far lower in attacks with guns than in attacks with knives and other attacks.

Even individuals trained and presumably emotionally prepared to shoot under stressful conditions, such as police officers, usually cannot hit their targets. Police shooting policies usually forbid firing warning shots, and thus when the officers fire their guns they intend to shoot suspects. Nevertheless, police officers were able to inflict one or more gunshot wounds on an adversary in only 37% of the incidents in which they intentionally fired at someone (Kleck, 1997). This success rate is probably even lower among civilians, who have not had the training and experience of police officers, and the NCVS data support this expectation. Thus, there is strong reason to believe that the net effect of offender gun use in violent crimes is that it decreases the fraction of attacks resulting in injury.


About 1 of 7 assaultive gunshot woundings known to the police results in death (Kleck, 1997). Because many less serious nonfatal gunshot woundings never come to the attention of authorities, the true death rate is almost certainly lower than this. Nevertheless, gunshot wounds are more likely to result in death than are those inflicted by a knife, the weapon that is generally assumed to be the next most lethal among those that could be used in the same circumstances as guns. Most police-based and medical studies indicate that gunshot woundings are about three to four times more likely than knife woundings to result in the victim’s death (Kleck, 1997).

One of the central mysteries of the guns-violence field is the degree to which the higher fatality rate of gunshot attacks is due to the greater inherent lethality of firearms or to the greater degree to which people who use guns with which to attack are more willing to kill their victims. In other words, is the difference in fatality rates due to differences in weapon lethality or differences in attacker lethality? Attackers do not randomly choose their weapons or merely use whatever is available. It is a rare gun homicide that occurs when a knife or blunt instrument is not also available, and all gun killers obviously also have hands and feet with which they could have attacked the victim. Thus, guns are chosen by aggressors over other available weapons. Furthermore, scholars generally agree that aggressors choose weapons suited to their goals and that the aggressors who choose guns probably have more lethal intentions than those who choose knives. Consequently, some of the higher fatality rates of gun attacks are due to attacker differences instead of weapon lethality differences. Unfortunately, unless one can somehow measure and control for attacker lethality in assaults, it is logically impossible to use data on assault fatality rates to separate the effects of a weapon’s technical properties from the closely associated effects of the attacker’s willingness to seriously hurt the victim.

The comparison of gun lethality versus knife lethality, however, is something of a red herring, or at the very least a distraction from more policy-relevant issues. The vast majority of existing gun laws and proposed control measures apply exclusively to, or with greater strictness toward, handguns, whereas long guns, such as shotguns and rifles, are left relatively unregulated. Thus, many offenders are free to substitute long guns when handgun-only controls deny them the preferred handgun. Most homicides are committed under circumstances in which it was not essential that a handgun be used (concealability or easy portability of the weapon was not essential), so the substitution issue that is most frequently relevant to debates over handgun controls is the substitution of long guns for handguns, not the substitution of knives for guns.

There is little doubt that long guns are more lethal than handguns. Shotguns fire more projectiles, and create more wounds, than handguns, whereas rifles fire bullets at a higher velocity, producing wounds with greater penetration into the victim’s body. Long guns are also more accurate than handguns; a shooter using a long gun is more likely to wound the victim. To the extent that handgun controls attain their proximate goal of denying handguns to at least some prospective attackers, but do not significantly restrict access to long guns, they are more likely to lead to substitution of more lethal weapons than less lethal ones. The policy implication is that if a subset of the population is to be legally denied guns, the restriction should cover all gun types, not just handguns.

Offender Gun Use in Robbery

Weapon effects in the context of robberies merits its own separate discussion. Gun effects may differ from those in assaults, because the robber’s primary goal is to obtain the victim’s property, and threats or use of force are largely tools for achieving that goal. About 25% of robberies involve offenders armed with guns, and about 5% of all homicides that occurred in 2006 were committed with guns and linked with robbery (computed on the basis of statistics from the U.S. Federal Bureau of Investigation, 2008a). The effects of offender possession and use of guns on the frequency and outcomes of robberies are quite complex, but research supports the following conclusions:

  • Total gun ownership levels (criminal and noncriminal combined) have no net effect on total robbery rates. On the other hand, we do not know the impact of gun ownership among criminals, or rates of gun carrying—and thus the immediate availability of guns for robbery—on robbery rates.
  • Higher gun ownership levels probably increase the rate of gun robberies, and decrease the rate of non-gun robberies, thereby increasing the fraction of robberies involving guns.
  • Injuries are less common in gun robberies than in non-gun robberies; therefore, decreases in gun use among robbers would probably increase the fraction of robberies that result in injury.
  • When injuries are inflicted on robbery victims, those inflicted by gun-armed robbers are no more likely to result in hospital treatment of some kind than those inflicted by other robbers. Injuries inflicted by gun-armed robbers are more likely to result in hospitalization overnight than those inflicted by unarmed robbers, but they are about the same in this respect as injuries in knife robberies and somewhat less likely to result in overnight hospitalization than injuries inflicted by robbers armed with weapons other than guns or knives. Thus, there is currently no empirical basis for believing that if knives were substituted for guns, the fraction of injuries requiring hospital treatment or overnight hospitalization would decrease.
  • Robbers armed with guns are more likely to obtain the victim’s valuables. This is partly due to the fact that victims are less likely to resist gun-armed robbers. Thus, if fewer robbers were armed with guns, more victims would probably manage to retain their property.
  • Guns enable robbers to tackle more lucrative and risky targets, such as businesses, instead of more vulnerable ones, such as women, children, and the elderly. Reducing gun availability could cause robbers to switch from the former to the latter targets, shifting the burden of robbery to those most vulnerable to injury and least able to bear the financial losses.
  • Gun robberies are more likely than nongun robberies to result in the death of the vulnerable victims. It is unknown, however, whether this is due to the lethality of guns or the greater willingness to kill of robbers who use guns. Gun reductions therefore may or may not produce any reduction in robbery murders, depending on the impact of gun scarcity on (a) the number of robberies; (b) how much of an increase in the number of injuries this causes; and (c) how much the fatality rate declines among this increased number of injuries, assuming it declines at all. The issue is further complicated by the fact that most gun control legislation restricts primarily or only handguns, but most incarcerated felons say they would substitute long guns, such as sawed-off shotguns, if they could not carry handguns. This suggests that laws that reduce only the availability of handguns would increase the fraction of robbery attacks resulting in death by inducing the substitution of more lethal long guns.

In sum, gun control policies that reduce gun possession among robbers would have the desirable effect of decreasing the rate at which robbers obtain their victims’ property, and they might or might not reduce the number of robbery victims killed. On the other hand, gun scarcity would also probably increase the number of robbery injuries and shift the burden of victimization to victims less able to bear the burden, without reducing the number of robberies and without necessarily reducing robbery killings. Therefore, it is unclear whether the overall set of social consequences of gun scarcity would be favorable with regard to robbery.

Crime-Disrupting Defensive Effects of Victim Use of Guns

Defensive gun use by crime victims is both common and effective in preventing injury to the victim and property loss. People who use guns during crime incidents are less likely to be injured or lose property than people who either adopt other resistance strategies or do not resist at all. These effects are usually produced without shooting the gun and are almost always produced without wounding or killing the criminal: Only 24% of gun defenders even fire the gun (including warning shots), only 16% try to shoot the perpetrator, and at most 8% wound the offender (evidence summarized in Kleck, 1997, Chap. 5).

Victims’ defensive use of guns almost never angers or otherwise provokes offenders into attacking and injuring the resisting victims. It is extremely rare that victim gun use is followed by injury to the victim, and some of these few injuries would have been inflicted anyway, regardless of victim resistance. In any case, it is clear that, regardless of whether victim gun use occasionally provokes offender aggression, the net effect of victim gun use is to reduce the likelihood that the offender will hurt the victim.

The largest, most nationally representative samples of crime incidents on which we have information about victim resistance strategies and their consequences are drawn from the NCVS, conducted by the U.S. Census Bureau. The most extensive analysis of these data was conducted by Tark and Kleck (2004). They found that, among 45 sample cases of victims who used a gun to attack the offender, none were injured after using the gun, and of 202 sample cases of victims who used a gun to threaten the offender, just 7.7% were injured after using the gun. They also found that victims who resisted with a gun typically did so under more dangerous and difficult circumstances than victims who used other strategies (e.g., when the victim faced offenders who were armed, when the victim was outnumbered by the criminals, or when the victim was already injured). If one takes into account these greater dangers, victim gun use appears to be even more effective in preventing injury than the already very low injury rates suggested.

The impression from earlier studies that victim resistance increases the odds of being injured appears to be the product of a simple research error: the failure to take account of which came first, victim resistance or injury. Crimes in which a resisting victim was injured turn out to consist almost entirely of incidents in which the victim resisted after the offender attacked and injured him or her (i.e., the injury provoked victim resistance; resistance did not provoke the offender to inflict injury) (Kleck, 1997).

Early pro-control propaganda often claimed that when victims attempt to use guns defensively, offenders often take the guns away from them and use them against the victim. This is false. The only significant factual foundation for this claim appears to be the fact that police officers are occasionally killed with their own guns. This phenomenon is, however, extremely rare (it happened just once in the United States in all of 2006) and not as relevant to the issue of defensive use of guns as it seems. From 1997 through 2006, an annual average of 4.8 police officers in the United States were killed with their own guns, out of a total of 665,555 full-time sworn officers in the nation. Furthermore, these extremely rare incidents typically do not involve the officer attempting to use the gun defensively; instead, they usually involve the suspect snatching the gun from the officer’s holster or stealing it from his or her vehicle. Thus, the officer’s gun was available to be obtained by the criminal suspect because the officer was not using the gun for self-protection (Kleck, 1997, pp. 168-169; U.S. Federal Bureau of Investigation, 2008b).

Deterrent Effects of Gun Ownership Among Potential Victims

Evidence also indicates that some criminals may be deterred from making criminal attempts in the first place by the possibility of victims using guns against them. Criminals interviewed in prison indicate that they have refrained from committing crimes because they believed a potential victim might have a gun (Kleck, 1997). Likewise, anecdotal evidence indicates that crime rates have dropped substantially after highly publicized instances of prospective victims arming themselves or being trained in gun use, or victims using guns against criminals. Evidence also supports the hypothesis that U.S. burglars are careful to avoid residences where the victims are home because they fear being shot. Whereas 43% of British residential burglaries are committed while victims are home, only 9% of residential burglaries in the United States are committed under such circumstances (research summarized in Kleck, 1997, Chap. 5). None of this evidence is strong or decisive; instead, one can say only that it is consistent with the hypothesis that criminals are deterred from attempting some crimes by the possibility of confronting a victim with a gun.

The Net Effect of Gun Ownership Levels on Crime Rates

The research on gun use by victims has yielded very consistent results: It reduces the likelihood of injury or property loss. Thus, gun possession among largely noncriminal prospective victims has beneficial effects. On the other hand, gun possession among criminals has a mixture of both harmful and inadvertently beneficial effects. Consequently, the net effect of overall gun ownership levels on violence rates is not self-evident on the basis of the research discussed earlier.

Dozens of studies of the effect of gun ownership levels on crime rates in macrolevel units such as cities and states have been conducted, but most of the research is seriously flawed. In particular, most studies have failed to properly model the possibility of a two-way relationship between violence rates and gun ownership rates, making it impossible to interpret the meaning of a positive association between the two (Kleck, 1997). Although more guns may lead to more crime, higher crime rates may motivate more people to acquire guns for self-protection. Likewise, most of these studies did not use measures of gun levels that are known to be valid—the researchers were actually measuring something other than gun ownership levels, making it impossible for them to assess the effect of gun levels.

Most of this research has found no effect of gun levels on rates of robbery or aggravated assault. The evidence on homicide rates, on the other hand, is much more mixed, although studies that have used validated measures of gun ownership and that addressed the possible two-way causal relationship mostly have found no net effect of gun levels on homicide rates (Kleck, 1997, Chap. 7). The most sophisticated recent research indicates that the net effect of overall gun ownership (both criminal and noncriminal gun ownership combined) on homicide is actually negative; that is, overall gun levels reducethe homicide rate, probably because the homicide-reducing effects of noncriminal gun ownership outweigh the homicide-increasing effects of criminal gun ownership (Kovandzic et al., 2005). Because this research relied on highly complex statistical procedures, however, one can be confident that these findings will be challenged.

How Do Criminals Acquire Guns?

Some scholars have asserted that professional gun traffickers (i.e., criminals who illegally sell substantial numbers of guns for profit) are significant sources of guns for criminals. The best available evidence, however, fails to support this claim. To be sure, burglars and other thieves sell guns that they come across in their criminal activities, but they average fewer than 4 guns a year, and some of these are sold to noncriminal buyers. Illicit gun selling is almost all done at a very low volume. Typical trafficking operations uncovered by law enforcement authorities handle fewer than 7 guns each, and the federal Bureau of Alcohol, Tobacco and Firearms uncovers fewer than 15 high-volume (> 250 guns) operations in the United States each year. High-volume trafficking probably supplies less than 1% of the guns in criminal hands.

Trafficking activity apparently has no measurable effect on levels of gun possession among criminals, or on violent crime rates. One likely explanation would be that nearly all traffickers’ potential criminal customers have other sources of guns (especially the pool of locally stolen guns) and are not dependent on traffickers. Consequently, even the bestdesigned strategies aimed at reducing gun trafficking are unlikely to have any measurable effect on gun possession among criminals or on violent crime rates (Kleck & Wang, in press).

Instead, theft appears to be crucial as the mechanism that brings guns into criminal hands. NCVS data indicate that at least 400,000 to 600,000 guns are stolen each year, a number many times higher than any evidence-based estimate of the volume of trafficked guns (Kleck & Wang, in press). As a result, at any one time there are millions of stolen guns circulating among criminals. The volume of gun theft is so large that, even if one could completely eliminate all voluntary transfers of guns to criminals, including either lawful or unlawful transfers, and involving either licensed dealers or private citizens, and even if police could confiscate all firearms from all criminals each year, a single year’s worth of gun theft alone would be more than sufficient to rearm all gun criminals and supply the entire set of guns needed to commit the current number of gun crimes (Kleck, 1997, pp. 90-94). Interviews with incarcerated felons indicate that most guns acquired by criminals were probably stolen at some time in the past (Wright & Rossi, 1986).

Most gun theft is a by-product of residential burglary and other thefts from private owners. Less than 2% of stolen guns are stolen from gun dealers. Criminals do not typically go out looking for guns to steal but instead steal those they happen to come across in the course of crimes, most commonly in thefts from homes or vehicles. They usually sell the guns they steal, but most gun thieves have also retained at least one gun in their careers for their own use. They typically do not keep the gun not because they did not already have one but because the stolen weapon was “a nice piece.” Thus, criminals most commonly use theft as a way of upgrading the quality of their weaponry instead of as a way of becoming armed (Wright & Rossi, 1986).

Wright and Rossi (1986, p. 185) found that 16% of the felons’ handguns had been purchased from retail, presumably licensed, sources, probably because the criminals did not have any disqualifying criminal convictions at the time of purchase or because no background checks were required at that time. Surveys of incarcerated criminals also indicate that offenders believe they could get guns from multiple types of sources, so eliminating a single channel of guns usually would not prevent acquisition of a gun.


The widespread availability of guns in America affects crimes in far more complicated and surprising ways than is generally known. Guns in the hands of crime victims have primarily violence-reducing effects, whereas guns in the hands of criminals have both violence-increasing effects and, more surprisingly, some violence-reducing effects as well. The implications for crime control policy are that gun control efforts should focus narrowly on depriving criminals from guns, because disarming victims and prospective victims would have predominantly crime-increasing effects. It would therefore be unwise to try to reduce gun availability among criminals by reducing it in the general population in the hope that this would reduce the flow of guns from noncriminals to criminals via theft. However, even the narrowly focused disarming of criminals will not necessarily have exclusively violence-reducing effects; criminal-centered gun control efforts will succeed only if the crime-increasing effects of guns in the hands of criminals are stronger than the crime-decreasing effects.

The control efforts most likely to minimize criminal gun use are those that operate most directly on the last links in the chain of possession of guns, just prior to a criminal using the gun to commit a violent crime. Thus, efforts to intercept guns while carried through public spaces on the way to a crime scene are more likely to be effective than efforts to restrict manufacture, importation, or retail sales of guns, because the causal chain resulting in criminal gun use is so much shorter and direct from gun carrying in public to use in a crime. Thus, one of the more promising approaches to reducing gun crime is improving the ability of police officers to detect concealed gun carrying and increasing their inclination to make arrests for unlawful carrying of firearms.