Security camera statistics span national estimates, local public-camera systems, school surveys and evaluations of crime outcomes. The figures below distinguish observed counts from forecasts, surveys, models and quasi-experiments, while retaining each study’s original geography and period.
Key Security Camera Statistics
The most useful headline figures are:
- 70 million surveillance cameras were installed in the United States in 2018, up from 47 million in 2015.
- About 85 million cameras were forecast for the United States in 2021, a pre-2021 estimate described as 21% above 2018.
- Seattle DOT operated 284 CCTV traffic cameras in June 2020.
- 90% of surveyed U.S. school districts reported security-camera use in the survey period reported in the 2003 final report.
- 78% of surveyed U.S. school districts fed camera monitors to viewers in real time in that same reported survey period.
- 35% of surveyed U.S. school districts used computer-based camera networks in the reported 2003 survey period.
- A 76-study international meta-analysis found a 14% reduction in the odds of overall crime for CCTV compared with controls across evaluations from 1978 to 2017.
- CCTV was associated with a 16% reduction in the odds of property offenses across 22 studies in evaluations from 1978 to 2017.
- CCTV was associated with a 16% reduction in the odds of vehicle crime across 23 studies in evaluations from 1978 to 2017.
- CCTV had no statistically significant effect on violent offenses across 29 studies in the same 1978–2017 evaluation set.
- Philadelphia camera viewsheds had a statistically significant 13.3% reduction in expected crime counts after implementation, compared with control areas, in the 2015-published evaluation.
- Baltimore’s Greenmount area had a statistically significant 15% decrease in aggregated crime after installation in the evaluation through 2008.
- Chicago’s Humboldt Park study area had about 53 cameras per square mile during the 2001–2006 evaluation.
- Seattle estimated CCTV installation or replacement at $20,000 per camera in a 2020 cost estimate.
- Seattle estimated 2021 CCTV operating costs at $1,680,872, including personnel and maintenance.
- 18 of 24 surveyed U.S. federal agencies reported facial-recognition use in fiscal year 2020.
Contents
- Key Security Camera Statistics
- How Many Security Cameras Are in Use?
- Security Camera Adoption in Schools and Public Institutions
- Crime and Clearance Outcomes from CCTV
- Camera Density, Diffusion and Displacement Data
- Security Camera Costs and Economic Measures
- Video Analytics Adoption and Governance Measures
How Many Security Cameras Are in Use?
The broadest U.S. figures are estimates of installed surveillance cameras, not a census of every residential, commercial, school or public device. The U.S. Bureau of Labor Statistics reported growth from 2015 to 2018, followed by a forecast for 2021.
| Period | United States camera estimate | Measurement type |
|---|---|---|
| 2015 | 47 million installed cameras | Estimate |
| 2018 | 70 million installed cameras | Estimate |
| 2021 | About 85 million cameras | Pre-2021 forecast |
Source: Investigation and security services. The 2021 figure was a forecast, not a later observed count.
Local transportation systems provide a narrower but concrete view of public-camera inventories. Seattle DOT operated 284 CCTV traffic cameras in June 2020, then replaced 14 and added 43 from July 2020 through December 2021.
Those Seattle figures describe one agency’s traffic-camera system, so they should not be added to the national estimates or treated as representative of all public cameras.
Security Camera Adoption in Schools and Public Institutions
A U.S. school-district survey reported in the 2003 final report found widespread camera use among its sample. The sample was unweighted and included 40 districts, with some individual measures answered by 39 districts; these are historical survey results, not current national prevalence estimates.
| Reported practice | Share of surveyed districts | Survey scope |
|---|---|---|
| Used security cameras | 90% | 40-district sample |
| Fed monitors to viewers in real time | 78% | 40-district sample |
| Used closed-circuit television systems | 82% | 39 districts answered |
| Used computer-based camera networks | 35% | 40-district sample |
| Installed interior cameras | 85% | 40-district sample |
| Installed exterior cameras | 80% | 40-district sample |
Source: Surveillance Tools for Safer Schools, Final Report. Among camera-using districts, 97% had installed cameras in high schools.
The survey also recorded the types of equipment and deployment choices in the sample. Color-image cameras were used by 58% of districts, pan-tilt-zoom cameras by 60%, wireless cameras by 21%, hidden cameras by 40%, false cameras by 15%, and bullet-resistant cameras by 5%.
These percentages describe equipment reported in the historical sample, not performance rankings. They also should not be read as mutually exclusive categories: a district could use more than one camera type or location.
Crime and Clearance Outcomes from CCTV
The strongest cross-study evidence in the supplied research comes from an international meta-analysis of CCTV evaluations conducted from 1978 through 2017. It found a 14% reduction in the odds of overall crime versus controls across 76 studies, while results varied by offense category.
| Outcome | Evidence base | Result |
|---|---|---|
| Overall crime | 76 studies | 14% lower odds |
| Property offenses | 22 studies | 16% lower odds |
| Vehicle crime | 23 studies | 16% lower odds |
| Violent offenses | 29 studies | No statistically significant effect |
Source: Practice Profile: Closed Circuit Television (CCTV) Surveillance. The pooled studies were heterogeneous, so these results summarize associations across different settings rather than a guaranteed effect for every installation.
Local evaluations show why geography and design matter. In Philadelphia, a camera program introduced in 2006 was associated with a statistically significant 13.3% reduction in expected crime counts in camera viewsheds after implementation compared with control areas; the profile reports a nine-month follow-up for most sites and no statistically significant impact on serious crime.
Baltimore’s Greenmount area recorded a statistically significant 15% decrease in aggregated crime after cameras were installed in August 2005, with a difference-in-differences estimate of 8.22 fewer crimes per month. In the Baltimore Tri-District, cameras installed in March 2006 were associated with a decline of 12.35 total crimes per month and a nearly 35% decrease in total crime in that local area.
The Tri-District analysis also estimated decreases of 1.54 inside-larceny incidents per month and 2.06 robbery incidents per month relative to controls. These are local quasi-experimental estimates, not national clearance-rate statistics.
Source: Evaluating the Use of Public Surveillance Cameras for Crime Control and Prevention and Program Profile: CCTV in Philadelphia (Pa.).
Camera Density, Diffusion and Displacement Data
Camera counts can look very different across small study areas. During a Chicago evaluation covering 2001–2006, Humboldt Park had about 53 cameras per square mile, while West Garfield Park had about 36 cameras per square mile; the intervention in both study areas occurred in August 2003.
Baltimore researchers also examined whether crime changes appeared to move outside a camera area. In the Tri-District, total-crime weighted displacement quotients were 0.59 at 500 feet and 0.50 at 1,000 feet over 12 months.
Under the study convention, the first 12-month comparison produced a -1.19 quotient for inside larceny and -0.76 for robbery at 1,000 feet; the source notes that negative values indicate displacement under that convention.
| Location or measure | Value | Period and scope |
|---|---|---|
| Chicago Humboldt Park | 53 cameras per square mile | 2001–2006 study area |
| Chicago West Garfield Park | 36 cameras per square mile | 2001–2006 study area |
| Baltimore Tri-District total-crime WDQ | 0.59 at 500 feet; 0.50 at 1,000 feet | August 2004–July 2006 |
| Washington, DC Mount Vernon Square cluster | 13 cameras | January 2005–February 2009; intervention October 2006 |
Source: Evaluating the Use of Public Surveillance Cameras for Crime Control and Prevention. The Washington, DC analysis found about 70 incidents per month within 200 feet of cameras before intervention, compared with about 69 incidents per month in the matched comparison area.
These measures are not interchangeable: cameras per square mile describe density, while weighted displacement quotients describe a study-specific relationship between intervention and nearby crime patterns.
Security Camera Costs and Economic Measures
Seattle’s public CCTV records show that operating costs extend beyond the camera hardware. The city estimated installation or replacement at $20,000 per camera in 2020, while annual operating estimates included personnel and maintenance.
| Seattle CCTV cost measure | Amount | Period |
|---|---|---|
| Maintenance | $242,268 | 2019 |
| Maintenance | $227,799 | January–June 2020 only |
| Operating cost, personnel and maintenance | $1,680,872 | 2021 estimate |
| Personnel | $1,414,574 | 2021 estimate |
| Maintenance | $259,969 | 2021 estimate |
Source: Seattle DOT’s Closed Circuit Television Technology Costs Totaling $1,078,982 Including Personnel and Maintenance in 2019. The installation figure was a manager estimate, and the 2020 maintenance amount covered only the first two quarters.
From July 2020 through December 2021, Seattle’s 14 replacements cost $128,185 in total, including $112,868 for cameras and $15,317 for labor. The 43 additions cost $440,759, including $346,666 for cameras and $94,093 for labor.
Economic evaluations may also assign societal values to offenses. A 2009-dollar Urban Institute model valued burglary at $13,171 per offense, larceny or theft at $7,353, robbery at $120,878, and murder or non-negligent homicide at $1,435,507; these are model-based societal costs, not camera-program spending.
The same public-surveillance evaluation reported Chicago study-area camera costs of approximately $6.8 million through August 2006, with some maintenance inputs using a Baltimore proxy. That limitation matters when comparing the Chicago cost figure with Seattle’s agency-specific records.
Video Analytics Adoption and Governance Measures
Facial-recognition statistics measure an analytic capability, not the number of cameras in an inventory. A Government Accountability Office survey of 24 U.S. federal agencies for fiscal year 2020 found that 18 agencies reported facial-recognition use, while the reported purposes varied.
| Reported facial-recognition use | Agencies | Population and period |
|---|---|---|
| Any reported use | 18 of 24 | U.S. federal agencies, FY2020 |
| Digital access or cybersecurity | 16 of 24 | U.S. federal agencies, FY2020 |
| Generate investigation leads | 6 of 24 | U.S. federal agencies, FY2020 |
| Physical security or facility access | 5 of 24 | U.S. federal agencies, FY2020 |
| Research and development | 10 of 24 | U.S. federal agencies, FY2020 |
Source: Facial Recognition Technology: Current and Planned Uses by Federal Agencies. The survey notes that facial recognition could involve still images or video, and the reported uses were agency self-reports; research and development does not necessarily mean deployment or live-camera use.
Taken together, these statistics show several distinct layers of security-camera measurement: installed-camera scale, institutional adoption, local density, evaluated crime outcomes, operating costs and analytics governance. The layers should remain separate when interpreting what a camera count or a reported percentage actually represents.