FBI Seal Crime in the United States, 2002 Department of Justice, Federal Bureau of Investigation
Contact Information Contents Page FBI Home Page Appendices Law Enforcement Personnel Special Report Persons Arrested Crime Index Offenses Cleared Crime Index Offenses Reported Summary of UCR Data

Special Report


Bank Robbery in the United States


According to the Uniform Crime Reporting (UCR) Program, robbery is the taking or attempting to take anything of value from the care, custody, or control of a person or persons by force or threat of force or violence and/or by putting the victim in fear. The focus of this study, bank robbery, is a subtype of robbery targeted at banks. Because of this element of force or the threat of force, bank robbery is highly feared among the population.1

Some view robbery in the context of violence; others maintain that robbery offenders come from a subculture of theft.2 Sometimes it is difficult to separate the two. The UCR Program classifies robbery as a crime against property and includes robbery in its violent crime total.

A bank robbery is indicated when the crime is robbery and the location is a financial institution. UCR-National Instant-Based Reporting System (NIBRS) standards state that the victims in a robbery can be either persons or entities, i.e., businesses, financial institutions, etc., or both.3 In a bank robbery, the primary victim is the bank itself, but the teller being threatened or injured is also a victim.

A computation of UCR Summary data showed that a bank robbery occurred just under every 52 minutes in 2001, accounting for 2.4 percent of all robbery in the United States.4 This represented a total loss of approximately $70 million. While this seems like a large amount of money taken, the average amount of money taken in a bank robbery over the period 1996 through 2000, according to NIBRS data is less than $5,000.

The crime of robbery showed a clearance rate of only 24.9 percent in 2001. The clearance for bank robbery was 57.7 percent in 2001.5 This is a relatively high clearance rate when compared with that of other Part I crimes.* Only murder, at 62.4 percent, has a higher percentage of crimes cleared by arrest.

Even with such a high clearance rate, bank robbery remains prevalent. Bank robbery has been the subject of many studies.6 Because of the number of incidents, the amount of money taken, and the fear engendered in the public, bank robbery is a serious problem in the United States. Dr. Yoshio Akiyama of the FBI addressed this question in 1983 in the Crime Indicators System, Fourth Semiannual Briefing on Crime. That study used a 10-year time series to show the prevalence and characteristics of bank robbery incidents, a profile of offenders, and an analysis of the length of time from the incident until clearance.

The present study will update and extend parts of that earlier study.


 The general objective of this study is to examine three different criminal justice databases maintained by the FBI. Their similarities and differences are pointed out and discussed with the purpose of producing a fuller picture of bank robbery than that created when using only one of these databases. A further and no less important objective is to provide some assessment of the NIBRS bank robbery data by comparing it with the Bank Crime Statistics database, even though the collection methods, the scope, and content of these databases are different.

To address these objectives, a time series from the Bank Crime Statistics (BCS), collected by the Violent Crimes/Fugitive Unit of the FBI, covering the period 1973 to 2001 was generated and compared to the time series for Summary UCR data and to NIBRS data. NIBRS data on bank robbery incidents used for this analysis is for 1996-2000. Although Summary data have been collected by the FBI since 1930, its comprehensiveness concerning bank robbery is limited. Therefore, only the crime counts and estimates from 1990 through 2001 were examined for comparison to BCS data from the same period.†

The study questions in this analysis are designed to compare and contrast the databases on the subject of bank robbery as it is reported to the FBI and are divided into two areas, characteristics of the incident and characteristics of the offender(s).

Further, this study will discuss the general compatibility of the Summary UCR data, the historical BCS database, and bank robbery incidents identified in the NIBRS. By using data from all three of these databases it will be possible to present a fuller picture of the crime of bank robbery in the United States and how it is reported.

Study Question 1—Characteristics of the Incident

The level of analysis here is the incident itself. Variables that describe the incident, such as the number of bank robbery incidents per year, the state, the region, the time of day and day of the week, the violence—deaths, injuries, hostages taken—and the type of weapons used are addressed in question 1.

Study Question 2—Offenders

Question 2 concerns the offender characteristics. What is the age, sex, and race of the offender (or offenders)? What is the average number of offenders per incident? What is the previous bank robbery experience of offenders?

Data and Methodology

Data for the study come from three sources.

The UCR Program's data collections for the years 1990 and 2001. (Summary data)

The UCR Program is a law enforcement initiative that gives an annual depiction of crime in the United States. It is a nationwide cooperative statistical effort of over 17,000 city, county, and state law enforcement agencies that voluntarily report data on crimes that have been reported to them. The FBI has collected Summary data since 1930 with little change in the type of data collected and disseminated. Today, law enforcement agencies active in the UCR Program cover approximately 93.4 percent of the population of the United States.

The UCR's NIBRS data from 1996-2000

The NIBRS is the redesigned, expanded version of the Summary UCR system. NIBRS data differ from Summary data in that the NIBRS contains data on each single incident and arrest. While the Summary data are individual counts of seven Part I crimes, NIBRS collects data on 22 crime categories. Incident, offense, victim, offender, and arrest data are collected on each incident reported by a law enforcement agency. NIBRS is a richer, disaggregated database than the Summary database that can be used to enhance law enforcement and crime research as well as strategic and administrative decision-making. A limited number of agencies began submitting NIBRS data to the FBI's UCR Program in January 1989.

The BCS data collected by the FBI from 1970-2001

In 1934 Congress enacted the Bank Robbery and Incidental Crimes Statute, making it a federal crime to rob any national bank or state member bank of the Federal Reserve System. This statute was expanded to include bank burglary and bank larceny and similar crimes committed against federally-insured savings and loan associations and Federal credit unions. The investigative jurisdiction under this statute has been delegated to the FBI, which today investigates a bank crime concurrently with local law enforcement.7

The Violent Crimes/Fugitive Unit of the FBI has collected descriptive data on bank robberies since 1970. This is a database that FBI Special Agents in the 56 field offices use when investigating bank robberies. The variables concern the incident, the solution, the mode of operation, and offender characteristics. Although these data are primarily meant to be used as an investigative tool to clear the particular crime, much of the data contained in the BCS can be used for quantitative research as well. These data can be used alone by the researcher or in concert with other statistical databases, specifically, the NIBRS database, to present a fuller rendering of the bank robbery incident.

The BCS database contains a more comprehensive representation of the U.S. population than the NIBRS database. It also includes several incident-level elements not included in the NIBRS. These are institution type, facility type, modus operandi, solution rates (analogous to clearance rates in the UCR Program definitions), types of security devices present in the incident, disguises used by the perpetrator(s), information on hostages that may have been taken, and the contents of any robbery notes.

Since the focus of BCS is a subset of all bank robberies collected by the UCR Program, care must be exercised when BCS statistics show deviations from those of Summary/NIBRS statistics. The definitions in the BCS data differ from those established by the UCR Program. It will be seen, however, that the two sets of statistics show striking similarities.


Frequency distributions and graphs are used to explore the consistencies and unique aspects of the databases to address the Study Questions.


Incident characteristics

Although the amount of money taken overall in any given year may seem high, approximately $70 million according to the BCS data for the period 1996 through 2000, the average amount netted from an individual bank robbery is less than $8,000 (BCS). The NIBRS data, covering less population than BCS, indicate an average of less than $5,000 per incident.

BCS reports the amount of money recovered is quite small. Over the period 1996-2000, $469,815,218.10 was reported as being taken and only $94,407,085.90 was recovered. This is only a 20 percent recovery rate.

Number of incidents

Table 5.1 shows the number of bank robbery incidents reported in the Summary UCR Program and in the BCS database of the FBI from 1990 through 2001. Figure 5.1 graphs the same data and makes the pattern easier to see. There was a substantial increase in the early 1990s, followed by an even more substantial decrease in the mid-1990s. Through 1999, the overall trend was down, but beginning in 2000 there was an upturn. The Summary UCR data always shows a greater number of incidents than the BCS database. There are two reasons for this. The first is measurement error, present in any data collection. The second reason may be because of the different, but overlapping, missions of the two databases. The Summary UCR number consists of all bank robbery incidents reported to the Program by local law enforcement. The BCS data includes bank robbery incidents reported to the Violent Crimes/Fugitive Unit of the FBI by the individual FBI field offices. Only in incidents where the FBI has investigative jurisdiction are the field offices required to collect and report data. FBI field offices do not report crime statistics to the UCR Program. Summary UCR data should contain these BCS incidents reported to UCR by the state or local law enforcement entity collaborating with the FBI on the investigation of the incident. Additionally, Summary data include incidents in which the FBI had no jurisdiction and, thus, no role. Therefore, the FBI became involved in the investigation of approximately 85 percent of all bank robbery incidents reported in the United States in 2000.

Although the time series of BCS is considered to reflect bank robbery trends in the Nation, the undulations in the number of bank robberies are also a result of the FBI involvements in the bank robbery investigations. The two time series track each other quite closely as we would expect. When one is moving downward, the other is moving downward as well, and when one turns up, so does the other.

Participation in the NIBRS has been more volatile over the period of the system's existence than either the Summary UCR or BCS participation over this same period of time. With only 17 percent of the U.S. population covered in 2000, the NIBRS reported 1,040 bank robberies. Summary UCR data showed approximately 8,565 bank robberies reported to the UCR Program by police agencies. The BCS recorded 7,310 bank robbery incidents with FBI involvement. A comparison of NIBRS data to Summary data to BCS data of this type is not meaningful. The Summary data and BCS data will drown out NIBRS data. However, other comparisons may be more fruitful between NIBRS and BCS data.

Regional breakdown

Table 5.2 shows a regional breakdown for NIBRS and BCS bank robberies in 2000. The UCR Program defines four regions in the United States and calls them Northeastern, Midwestern, Southern, and Western. The BCS also places states into four regions called Northeast, North Central, South, and West. The states in the UCR regions Northeastern, Southern, and Western are placed in the BCS regions Northeast, South, and West, respectively. The BCS region, North Central, contains the states that the UCR Program defines as Midwestern.

The percentages of bank robberies within the regions correspond somewhat between the two databases. Monotonically, they track from a low in the Northeastern, then the Midwestern, or North Central, through the South. The Western, or West region, is the odd one here, with the NIBRS showing it with only 11.3 percent of the bank robberies in 2000, while the BCS shows it with 33.3 percent. These disparities are due to the absence of major cities' participation in the NIBRS. It may also be that even though the Western region contains 13 states, only three of these report NIBRS data. Further, at least two of these three, Idaho and Utah, have small populations and thus would be expected to have fewer bank robberies. Moreover, California alone has more than one-half of the bank robberies in the entire 13-state region reported to BCS. California's 1,291 bank robbery incidents in 2000 are more than twice its closest competitor, Florida, and more than the entire Northeastern region. California drives the numbers in the Western region but is not represented in the NIBRS. On the other hand, 9 of 16 (17 when the District of Columbia is included) states in the Southern [South] region report NIBRS data. Further, 8 of 12 states in the Midwestern [North Central] region are NIBRS states.

Day of the week

Table 5.3 and Figure 5.2 show different presentations of the same data—bank robbery incidents by day of the week. The data are presented as percentages, i.e., the percentage of bank robberies reported in the NIBRS that happen on Sunday, on Monday, etc., and the same for the BCS data. In this way we can begin to make some comparisons between the two databases even though the difference in the absolute number of bank robberies in the two databases is quite high. The striking finding here is how closely the data in the two programs coincide. In both data series, Friday is the day on which most bank robbery incidents occur. Substantively, this may be because Friday has historically been payday for much of the United States and, thus, has required large deliveries of cash to branch banks. This may still be the case even in the modern world of electronic banking with direct deposit of paychecks and bill-paying either as an automatic withdrawal, by posted check, or over the Internet.

The second most prevalent days are Monday and Tuesday. The NIBRS reports a few more incidents on Monday than on Tuesday and BCS reports the opposite. Still, these differences are negligible and speak well for the integrity of the NIBRS data.

Most bank robberies from 1996 through 2000 happened on workdays, Monday through Friday, with very few occurring on the weekend. The NIBRS reports that workdays accounted for 89.83 percent of bank robbery incidents, while in the BCS the percentage was 93.85 percent for the period.

Time of day

Both the NIBRS data and the BCS data show that the time period during which most bank robberies occur is 9 a.m. until 11 a.m. Table 5.4 and Figure 5.3 present the time of occurrence of bank robberies reported in the NIBRS and BCS from 1996 through 2000 as a percentage of bank robbery incidents reported. The prominent detail presented here is the clear similarity of the two databases on this variable.

Weapons, violence, injury, and other crimes

One obvious reason for an individual to engage in bank robbery is economic where the motive is to obtain money. Another is that because of the low amounts of cash actually stolen per robbery, bank robbers are interested in projecting a persona of violence.8 Whichever is the case, the threat of violence is always present. Information on weapons used in the commission of a bank robbery, violence, injuries sustained, and other crimes is contained in the NIBRS data as well as the BCS data. The percentage involving an actual shooting reported in BCS is around 2 percent. Table 5.5 shows this percentage over the 1996–2000 period. BCS data displayed in Table 5.6 show that over this period, a firearm was present in about 32 percent of all bank robbery incidents. In almost all of those cases, 30 percent overall, that firearm was a handgun.

Table 5.7 presents NIBRS firearms data. Over the period 1996–2000, NIBRS reports firearms (including handguns) use in 49 percent of the 3,029 bank robbery incidents reported. Handguns were used in 38.5 percent of NIBRS incidents in which a firearm was used over the period.

It may be surprising that only between one-third and one-half of bank robbery incidents involve firearms. The perception one would tend to get from television or the movies is that a bank robber would never attempt a holdup without a firearm—and the more the better.

Table 5.8 holds another surprise. The incidence of violence and injury is very low. NIBRS data show that violence occurred in only 2.34 percent of incidents and BCS shows 4.84 percent over the time period. Given the low rates of violence, it should not be unexpected that the injury levels displayed in Table 5.8 are also quite low—5.58 percent for NIBRS data and 2.00 percent for BCS.

Regarding other crimes present in the incident, murder is very low at less than 1.0 percent in both databases, as are kidnapping and hostage-taking. Both NIBRS data and BCS data show that kidnapping/hostage-taking occurs in less than 2.0 percent of reported bank robberies.

Overall, the percentages in the table are close with neither database showing wildly divergent numbers; however, the numbers are so small for the NIBRS that we cannot take total comfort in the only-slight discrepancies the two databases show on these variables.

Offender characteristics

Despite what may be the popular perception, most bank robbery incidents, 79.9 percent in the NIBRS data over the period 1996–2000, were carried out by only one offender. Another 15 percent involved two offenders. Thus, over 95 percent of all the bank robbery incidents reported were attempted by two or fewer offenders.


Bank robbery offenders may not be as many or as varied as one might at first think. Using NIBRS and BCS data, we can analyze their race, and sex, and using NIBRS data we can examine age. Figure 5.4 shows the race of bank robbers from NIBRS data and BCS from 1996 through 2000 as a percentage of all offenders. There are similar patterns evident in the figure. Whites account for between 35 and 45 percent of all offenders in each of the years. Both NIBRS and BCS data bear this out and overall show the same level. Black offenders are responsible for between 45 percent and 55 percent over the period. If we average offenders by race over the 5 years, there is virtually no difference with whites averaging 40.84 percent in the NIBRS data and 39.45 percent in the BCS data. Similarly, the percentage of black offenders in the NIBRS data is very close to that in the BCS at 50.14 percent and 50.26 percent, respectively.


There is a great disparity between the number of male bank robbery offenders and the number of female offenders in both the NIBRS and the BCS databases. However, there is very little discrepancy when comparing the percentage of male offenders in NIBRS data to that in BCS and when comparing the number of female offenders. Figure 5.5 shows both of these comparisons. Male offenders are shown in dark red (NIBRS) and light red (BCS) and female offenders are shown in either black (NIBRS) or gray (BCS). In both databases, over 95 percent of the offenders are males, and less than 5 percent are females.

The percentage of male offenders in both NIBRS data and BCS is virtually the same. Table 5.9 shows the percentages of offenders that are identified as male in NIBRS data and BCS as well as the percentage identified as females. There is a strong correspondence between the two databases here.


Figure 5.6 displays the age and gender of offenders reported to the NIBRS from 1996–2000. The same information is contained in Table 5.10. Nearly 20 percent of all offenders are male, between the ages of 18 and 24. Males, aged 25–29 account for another 14 percent. Summing the two groups, we see that one-third of all bank robbery offenders are between 18 and 29 years of age. This is all the more astonishing because there are 703 offenders contained in the denominator that are either unknown or listed as missing data. If we drop the unknown and missing data from the denominator and recalculate the percentage, we find that 41.7 percent of bank robbery offenders reported in NIBRS data are 18–29-year-old males.

An examination of only male offenders shows these two age groups account for more then 45 percent of all male bank robbery offenders. Figure 5.6 shows a clear pulse in the late teens and early twenties that damps down in every subsequent age group.

Females show the same general pattern except that the numbers of female bank robbery offenders is much smaller than that of males.

Age, race, and sex are combined and presented in Table 5.11. The same patterns are visible in this table as shown earlier and separately. There are more males than females in every age group. There are more black males than white males in younger age groups and more white males than black in older (>35) age groups. There are more white females than black females. The number of Asian/Pacific Islanders and American Indians/Alaskan Natives are presented but are too small to analyze.

Prior bank robbery convictions

In Crime Indicators System, Fourth Semiannual Briefing on Crime (1983), Akiyama discussed the bank robber classifications of "professional" and "amateur." His discussion was based on a previous FBI report from 1977 that divided bank robbers into these categories. A "professional" in this classification scheme is a bank robber with a prior criminal record, despite his or her lack of success as evidenced by his/her incarceration. This professional is a bank robbery specialist. The "amateur" bank robber is a bank robber with no prior record. The amateur is presented as acting almost on a whim. The bank robbery to the amateur is almost a spur-of-the-moment undertaking with the robber engaging in very little planning. This individual robs banks to get the means to fulfill some more fundamental need, such as the need for drugs. Table 5.12 displays BCS data concerning the number and percent of subjects taken into custody for bank robbery who already have a conviction for bank robbery, bank burglary, bank larceny, or bank extortion. From 1996–2000, the average percent of "professional" bank robbers is 20 percent. This was more than the average in the earlier period from 1978–1982. Over that time period the average percent of "professional" bank robbers was 14 percent. This is still a clear indication that the great majority of bank robbers are amateurs and have not been convicted of a bank crime in the past.


There are several limitations to this study. Although Summary data have been collected by the FBI since 1930 and cover virtually the entire population of the United States, their comprehensiveness concerning bank robbery is limited. The only information available is the number of bank robberies, the percent of total robberies that that number represents, region of occurrence, bank robberies by population group, month of occurrence, and the amount of money taken in the aggregate. It is not possible to disaggregate Summary data to the individual incident.

Some bank robberies may not be captured in the database because of the Hierarchy Rule that limits reporting of only that crime in the incident that is highest in the "hierarchy" of Part I crimes as defined by the UCR Program. Both murder and rape are higher on this ordering of crimes than robbery. Therefore, if a bank robbery included a murder, the only crime entered into the Summary database is the murder. This would also be the case for a rape occurring within the bank robbery incident. Only the rape would be recorded and the bank robbery would be lost information.

Further, the bank robbery totals are collected on the form entitled Supplement to the Monthly Return of Offenses Known To The Police (Return A), but not on the Return A itself. If the supplement is not submitted, a robbery on the Return A cannot be counted as bank robbery. Thus, it may be the case that some robberies listed on the Return A and, therefore, in Crime in the United States, are bank robberies and are not captured in the Summary data.

Even though the NIBRS has distinct benefits as a data source, it is limited in its scope. Currently, agencies from 24 states, representing 17 percent of the U.S. population, participate in the program. These data lack the cross-sectional representation of incidents and cannot be treated as a sample. There are no cities participating that have populations greater than 1 million inhabitants. There are only 11 cities or consolidated counties that contribute NIBRS data whose population is more than 250,000. With this limitation, NIBRS data may not represent the crime experience in the entire United States.

Like the Summary UCR and NIBRS data, the BCS database also has its limitations. Only robberies of banks and financial institutions covered under the Bank Robbery and Incidental Crime Statute and its progeny are included. Further, the NIBRS includes, but BCS does not, specific information on each victim, offender, and arrestee. Finally, BCS is an investigative system; consequently the data are not available for use by the public.

Summary and Conclusions

The objective of depicting bank robbery from the data collected by the FBI has been met. Further, this realization of bank robbery through the use of these data has been an opportunity to compare and contrast elements in the databases—particularly the NIBRS data and the BCS data. These are preliminary findings and require further study.

The presentation of incident characteristics has emphasized the similarity of data submitted to the FBI's NIBRS program by local and state law enforcement to that submitted to the Violent Crimes/Fugitive Unit of the FBI by the separate FBI field offices. Both Summary UCR data and BCS indicate the same trends in the numbers of bank robberies over a 12-year period, 1990–2001.

Further, NIBRS data showing days of the week on which the greatest number of bank robberies occur and the hours during which they are most prevalent are very similar to BCS data, with Friday mornings generally the modal day and time for bank robberies.

In both databases, violence and injury are very low, an unexpected finding since one element of the crime is force or the threat of force. The similarity between the two databases concerning this unanticipated result adds further validation to the quality of NIBRS data.

Evidence of offender race and gender is also quite comparable between the two databases, with the number of whites committing bank robbery reported in NIBRS data very close to the number reported in BCS and the same for blacks. Reported gender of bank robbers is virtually identical in both databases.

Presenting the age data reported in NIBRS shows that a plurality of bank robberies are committed by offenders between 18 and 30 years of age.

Offenders are clearly amateurs and not bank robbery specialists as evidenced by the low number with previous convictions for a bank crime shown in the BCS statistics. That bank robberies do not involve the meticulously planned caper carried out by a group of highly experienced criminals is further borne out by the significant number of incidents involving only one or two offenders. NIBRS and BCS data show that the money obtained in a bank robbery is low, especially considering the amount of physical risk and the high probability of apprehension involved for the offender.

The money recovered is also not a very high percentage of that stolen. Both databases bear this out. This indicates that Akiyama (1983) was correct in his conclusion that most of these amateur bank robbers committed the crime to fulfill some more immediate need. More research is required, particularly into the aspect of drugs associated with this crime.

These findings are interesting and have significant implications for policymakers. This study and other research, such as which banks are most likely to be robbed, and which are more likely to be robbed more than once, in addition to spatial analyses adding variables such as location of the bank relative to escape routes, entrances to freeways, traffic patterns, location of nearest police station, etc., will allow law enforcement policymakers to develop better, more effective strategies for use in dealing with bank robberies.

The present study is also good news for the NIBRS program. The NIBRS has only 24 states that participate covering 17 percent of the population. Nevertheless, the percentages on the NIBRS variables examined here clearly accord with the percentages reported in the BCS. This should assure those who do not yet participate in the NIBRS program that they may reap large benefits from becoming a contributor to the Program.

Finally, since 9/11 the government has realized that information-sharing is a powerful tool with which to fight lawlessness. Databases such as those examined here should be examined to derive the maximum information toward this end.