False alarm statistics for San Diego

The measurement

How this can be counted directly

The false alarm figure quoted across the security industry is usually borrowed, undated, and traced to nobody in particular. It does not have to be. San Diego publishes every dispatched call for service with a call type and a disposition code, and it publishes the code dictionaries alongside. Call type 459A is "BURGLARY ALARM". Count the dispositions and you have a direct measure of what those dispatches actually produced, for the City of San Diego, year by year.

We counted a call as producing no crime outcome where the disposition was K or KB (no report required), U (unfounded), X (dispatched alarm call cancelled before a unit was on scene), W (no dispatch, or the incident cancelled before a unit was dispatched), CAN (cancel), or a duplicate-call code. We counted it as producing a crime outcome where the disposition was R, RB or RHR (report taken), A, AB or AHR (arrest made), or S (stolen or recovered vehicle). Those definitions come from SDPD's own published disposition dictionary, which is linked below.

One handling defect worth knowing about. The city's published 2024 calls-for-service file contains 970,649 rows but only 485,328 unique records: every row is duplicated. The 2025 file contains 12,227 duplicates. Every figure on this page is computed on de-duplicated records. Anyone reading the 2024 file straight will report roughly double the true alarm volume, and we mention it because a number you cannot reproduce is not much use to you.

SDPD burglary alarm calls and what they produced

Dispatched calls of type 459A, by outcome. Counts of calls, not of alarm activations: activations resolved by a monitoring centre never reach SDPD and are not in this data.

Source: San Diego Police Department, Police Calls for Service, City of San Diego Open Data Portal: data.sandiego.gov/datasets/police-calls-for-service. Disposition code definitions: pd_dispo_codes_datasd.csv. Period: calendar years 2019–2025. Type: COUNT of dispatched calls; percentages computed. Retrieved: 2 September 2026.

Derived from the above: the no-crime share has sat between 97.24% and 98.01% in every year since 2019. Across all alarm types in 2025 (burglary 17,075, robbery 971, panic 918, audible vehicle 305), alarm calls were 4.24% of SDPD's 454,752 de-duplicated calls for service.

Methodology

Limits of this analysis

Stated in full, because a derived statistic without its method is just a number somebody asserted.

This is a derived measure

The percentages are computed by us from SDPD's published disposition codes. They are not an official SDPD false-alarm statistic, and "no crime outcome" is not the same as the administrative false-alarm determination used for permits and fines. Cite it as our computation from SDPD data, which is how it is described throughout this page.

Calls, not activations

These are calls dispatched by SDPD. An activation that a monitoring centre resolves (by calling the keyholder, or by looking at video) never becomes a call for service and is not counted here. The true number of alarm activations in San Diego is larger and is not published by anyone.

The 2024 file is duplicated

The city's 2024 calls-for-service CSV contains every row twice: 970,649 rows, 485,328 unique records. The 2025 file carries 12,227 duplicates. All figures here are de-duplicated on the full record before counting, and any refresh of this page must do the same.

The DOJ guide is old

False Burglar Alarms, 2nd Edition was published in 2007 and its data is from roughly 2002 to 2005. It is quoted here because it is the DOJ's own published position and because our independent 2025 San Diego computation lands inside its range, not because it is current. Always attach the date.

Commercial and residential are not separated

The calls-for-service dataset does not record whether an alarm call was at a commercial or a residential address, so none of the San Diego figures on this page can be split that way. The DOJ's finding that commercial rates run higher is a national estimate, not a San Diego measurement.

Some incident types are excluded

SDPD states that the calls-for-service dataset excludes sensitive incidents including domestic violence, child abuse, suicide, sex crimes and stalking. That exclusion does not affect alarm call types materially, but it means the total calls-for-service denominator is not the whole of SDPD's workload.

Detail

What the 2025 dispositions actually say

The aggregate is less interesting than the breakdown. Most of these calls are not resolved by finding nothing at the premises: a large share never reach the premises at all, because they are cancelled in transit once somebody confirms there is nothing happening.

Source: San Diego Police Department, Police Calls for Service and the published disposition dictionary: data.sandiego.gov/datasets/police-calls-for-service. Period: calendar year 2025. Type: COUNT, de-duplicated. Retrieved: 2 September 2026. A further 231 calls closed as "other", 60 as duplicates and 25 with no disposition recorded; those are counted in the totals above but are not shown as separate rows here.

97.97% is our measure, not SDPD's administrative determination.

"No crime outcome" is a category we defined from SDPD's published disposition codes for the purpose of this page. It is not SDPD's administrative false-alarm determination, which is what governs permits and fines and is decided on different criteria. What the data supports is that 97.97% of dispatched burglary alarm calls in 2025 ended with no crime report and no arrest. "SDPD says 98% of alarms are false" is not a sentence we will write, because SDPD did not say it.

Corroboration

What the Department of Justice found nationally

Our San Diego figure is not an outlier. The DOJ reached the same range two decades ago, from entirely different data, and its guide is not flattering to the alarm industry, which is why it is worth citing.

The national false alarm rate

"The vast majority of alarm calls — between 94 and 98 percent (higher in some jurisdictions) — are false." ESTIMATE. Our own San Diego computation for 2019 to 2025 falls at the top of that range, at 97.24% to 98.01%.

The scale of it

"In the United States in 2002, police responded to approximately 36 million alarm activations, at an estimated annual cost of $1.8 billion." Nationally, false alarms accounted for between 10 and 25 percent of all calls to police. ESTIMATE, 2002 data.

Commercial premises are worse

Commercial properties "tend to have even higher false alarm rates than residential properties", because more people share responsibility for arming and disarming and the systems are more complex: as much as three times the residential rate. ESTIMATE.

The cost per call

"Each false alarm requires approximately 20 minutes of police time, usually for two officers." ESTIMATE. That is the mechanism by which an unverified alarm estate consumes response capacity that is not then available for anything else.

Alarms do not solve burglaries

The same guide notes that the clearance rate for United States burglaries "has remained below 15 percent for many years", and that the available research does not support alarms' value in catching burglars. We include this because it cuts against our own industry.

Unverified response is not recommended

The guide lists "providing an emergency police response to unverified burglar alarm calls" under responses it does not recommend. That is the DOJ's position on unverified alarms, and it is the reason this page is about verification, not about alarms.

Source: United States Department of Justice, Office of Community Oriented Policing Services, False Burglar Alarms, 2nd Edition, Problem-Oriented Guides for Police No. 5, by Rana Sampson: popcenter.asu.edu, catalogue record at ojp.gov. Period: guide published 2007; underlying data approximately 2001–2005. Type: ESTIMATE. Retrieved: 2 September 2026. Date this source every time you quote it: the 36 million activations and the $1.8 billion are 2002 figures and are not current.

The other half

And when the alarm is real, very little is solved

The false alarm problem is only half the picture. The other half is what happens on the small proportion of occasions when something did happen. In 2025 San Diego Police Department cleared 1,456 of 21,897 reported property crimes: 6.6%, about one in fifteen. For burglary specifically the SDPD clearance rate was 14.9%. Nationally, the FBI reports 17.4% of reported property crimes cleared in 2025.

Source: Federal Bureau of Investigation, Crime Data Explorer, offences and clearances, agency ORI CA0371100: cde.ucr.cjis.gov/LATEST/webapp; national figure stated directly in the FBI's UCR Summary of Reported Crimes in the Nation, 2025: cde.ucr.cjis.gov. Period: calendar year 2025. Type: COUNT; ratio computed. Retrieved: 2 September 2026. Cleared means cleared by arrest or exceptional means, not convicted, and clearances in a year may relate to earlier offences.

Put the two together and the position is uncomfortable but clear. An unverified alarm estate generates a very large number of dispatches that produce nothing, and when a genuine offence does occur the odds of it being cleared are low. Neither of those is fixed by a louder siren. What changes the picture is somebody looking at the site before a decision is made, and a recorded view good enough to be worth something afterwards.

That is what remote video monitoring does: an operator looks at the camera covering the sensor that triggered, decides whether anything is actually happening, and escalates or stands down on the basis of what is on screen. Where a site has an intrusion panel that should be watched around the clock, the same principle applies through commercial alarm monitoring. Both sit under our BSIS Alarm Company Operator licence, ACO #7501.

We looked for a source on how much verification cuts false dispatches. There isn't one.

We will not tell you verification cuts false dispatches by some particular percentage, because we found no primary source for a figure like that and are not going to invent one or borrow somebody else's. What the sources above support is narrower and still worth acting on: almost all unverified alarm dispatches produce nothing, the DOJ does not recommend responding to unverified activations, and verification is the mechanism that puts a human decision between the sensor and the dispatch. Whether it works on your site depends on where the cameras are and what they can actually see, which is a survey question.

What actually reduces a commercial site's false dispatches

The DOJ guide attributes chronic false activations mainly to user error, inadequate staff training, poor installation, and unsuitable or ageing equipment, not to the alarm concept itself. On a commercial site that usually resolves into a small number of unglamorous things: too many people sharing one code, a detector pointed at something that moves, a roller door that does not seal, a device that has been faulty for months and is worked around instead of fixed, and nobody owning the arming routine at close of business.

None of that is fixed by buying a different panel. It is fixed by walking the site, finding what is actually triggering, and correcting it, then putting verification in front of whatever remains. Where an existing system is generating dispatches, that is a service and repair job before it is a monitoring job, and it is handled under CCTV service and repair.

Fire alarm systems, and anything residential, are outside what we do.

Our monitoring scope is intrusion, video and access only. We do not hold a C-10 or a C-16, so we do not work on fire alarm systems; anything falling under a fire contractor's licence is coordinated with your own licensed fire alarm contractor, not performed or monitored by us. And we work on commercial premises only, not on private residences.

On permits: an SDPD alarm permit is required before a system is monitored in the City of San Diego, and it is renewed annually. The renewal is the part that quietly lapses. We will tell you what applies to your site at survey, not leave you to discover it.

The raw calls-for-service files, the disposition dictionary and the de-duplication script behind every figure on this page are retained. If you want to reproduce the counts, or you think we have read a code wrongly, tell us and we will correct the page.