How do the FBI, CDC and independent databases differ in counting and reporting homicides?
Executive summary
The Centers for Disease Control and Prevention (CDC) tallies "homicides" from death certificates completed by coroners and medical examiners and therefore captures deaths recorded as caused by another person, while the FBI aggregates "murders" reported to law enforcement through voluntary agency submissions to its Uniform Crime Reporting (UCR) system and Supplementary Homicide Reports (SHR), producing systematically lower counts in many years [1] [2] [3]. Independent databases — including academic reconstructions, media-collected open-source projects and the CDC’s National Violent Death Reporting System (NVDRS) or private aggregators like the Gun Violence Archive — use different inputs and cross-checking methods and therefore often sit between or alongside the government figures rather than perfectly matching either [4] [5] [6].
1. How the CDC counts homicides: death certificates and mandatory vital statistics
The CDC’s National Vital Statistics System classifies homicides using standardized death certificate coding filled out by coroners or medical examiners, producing a near-complete, mandatory record of deaths attributed to interpersonal violence that includes cases regardless of whether police later classify them as criminal or justified, and this approach is why many analysts favor CDC counts for completeness and time consistency [2] [1] [7].
2. How the FBI counts murders: voluntary police reports and SHR/NIBRS
The FBI’s measure comes from incidents reported by local law enforcement through the UCR program and, more granularly, the Supplementary Homicide Reports or the newer National Incident-Based Reporting System (NIBRS); participation and completeness are voluntary, some agencies fail to file supplemental reports, and methodological transitions (such as moving to NIBRS) and estimation choices can introduce gaps or timing issues that typically make FBI tallies lower than CDC totals [3] [8] [9].
3. Independent databases: triangulation, speed, and uneven coverage
Independent trackers — academic reconstructions, the NVDRS integration of coroners/police/death certificates, and open-source projects like the Gun Violence Archive — combine sources to identify events faster or include contextual variables missing from single-government sources; NVDRS, for example, has produced more complete counts of law-enforcement homicides than either the NVSS or FBI SHR in validation studies, while media and open-source databases can capture situational detail but may have geographic or selection biases [4] [5] [6].
4. Why the counts diverge: scope, source, timing, and classification
Differences arise because the CDC records deaths by cause on death certificates (capturing noncriminal scenarios and deaths not reported to police), the FBI records crimes known to police (excluding some deaths or federal/tribal incidents and depending on voluntary reporting), independent projects may exclude or include different event types or apply reconciliation rules, and transitions like the FBI’s NIBRS adoption or delays in death-certificate processing create year-to-year discrepancies and revisions [1] [3] [8] [10].
5. What each system is best for — and their blind spots
CDC data are strongest for comprehensive mortality trends and cross-jurisdictional comparisons because reporting is mandatory and medically coded, but they lack offender details and immediate incident context; FBI data are valuable for linking victim and offender characteristics and for policing-focused analyses at the department level but suffer from voluntary reporting gaps and lagged corrections; independent datasets can fill timing and context needs and sometimes correct government undercounts but vary in methodology and are vulnerable to selection and verification biases [2] [11] [6].
6. Practical implications for journalists, policymakers and researchers
Anyone using homicide figures must align their question to the data: use CDC counts to measure total lethal incidents by cause, FBI data to study crimes reported to police and offender-victim relationships, and independent/NVDRS resources for richer case-level narratives or quicker signals — and always disclose the data source because differences are real, persistent, and driven by reporting rules, voluntary submission, and coding practices rather than simple error [1] [3] [4] [12].