Baltimore entered 2026 with a figure that would have seemed improbable several years ago: 133 homicides in 2025, a 31.4 percent decline from the prior year and the city’s lowest annual total on record. In New York City, the first half of 2026 brought the fewest shootings, shooting victims, and murders recorded for any comparable six-month period. These are not merely statistical abstractions. For residents who have learned to distinguish fireworks from gunfire, for parents deciding whether a child can walk to the corner store, and for outreach workers who have attended too many funerals, a sustained decline can change the felt geography of everyday life.
The national picture is also encouraging, but more complicated than a victory headline. The Council on Criminal Justice’s Mid-Year 2026 Update examined 40 large U.S. cities. Among the 30 cities with usable homicide data, homicides were 18 percent lower than in the first half of 2025, representing 215 fewer deaths. Gun assaults declined 6 percent, robberies 17 percent, and carjackings 47 percent. Yet domestic violence incidents increased 8 percent, sexual assaults increased 3 percent, and drug offenses increased 12 percent. Homicide fell in 21 of the 30 reporting cities, meaning that the improvement was broad but not universal.
The first academic lesson is methodological humility. These figures are descriptive, not experimental. They tell us what changed in the reporting cities; they do not, by themselves, identify why. City samples change, police reporting practices differ, and short-term comparisons can be affected by unusually high or low baseline years. A falling citywide rate can also obscure neighborhood concentration, racial disparities, and the continuing burden borne by a relatively small number of blocks and social networks. The Council on Criminal Justice therefore warns against assigning the decline to a single cause.
A useful theoretical lens is collective efficacy, developed by sociologists Robert Sampson, Stephen Raudenbush, and Felton Earls. Collective efficacy combines social cohesion with a shared willingness to intervene for the common good. It helps explain why safety is not produced by law enforcement alone. A block becomes more capable of preventing harm when residents trust one another, institutions respond reliably, young people have credible adults to turn to, and neighbors believe that intervention will not expose them to retaliation or abandonment. This theory should never be used to blame communities. Concentrated disadvantage, residential instability, disinvestment, and institutional distrust can make collective action far more difficult.
Baltimore’s emerging model illustrates this combination of immediate risk reduction and community capacity. Safe Streets deploys credible messengers to mediate conflicts. Group Violence Reduction Strategy work concentrates attention on people at highest risk while pairing accountability with services. Hospital-based violence intervention reaches survivors during a period when retaliation or reinjury may be most likely. The city’s 2026 legislative testimony also acknowledges an important limit: although these strategies coincide with falling violence, existing analysis cannot attribute the decline to any single program. That candor strengthens, rather than weakens, the case for evaluation.
New York’s record-low first half similarly raises a practical question: can gains be maintained while public safety broadens beyond policing? The city’s new community-safety structure is intended to coordinate non-police responses involving mental health and other social needs. Its success should be assessed through outcomes that residents can recognize, not only through institutional activity counts.
A serious evaluation agenda would ask: Which neighborhoods experienced the decline? Did nonfatal shootings, retaliatory incidents, and victimization among young people fall as well? How quickly did outreach teams respond to conflicts? Were participants connected to stable housing, behavioral-health care, education, or employment? Did survivors of domestic and sexual violence receive safer and more accessible support? Results should be disaggregated by neighborhood, age, race, gender, and victimization type, with privacy protections and community oversight.
The central conclusion is neither complacency nor cynicism. Violence is not inevitable, and lower homicide counts are meaningful. But sustainability depends on treating safety as a social capacity: the ability of residents, institutions, service providers, and justice agencies to prevent harm, respond fairly, and learn from evidence. A decline is a foundation. The work is to make it durable—and to ensure that every neighborhood can feel what the citywide numbers appear to promise.
Sources and further reading:
Council on Criminal Justice, “Crime Trends in U.S. Cities: Mid-Year 2026 Update”: https://counciloncj.org/crime-trends-in-u-s-cities-mid-year-2026-update/
U.S. Department of Justice, “Partnerships Help Drive Sharp Decline in Violent Crime in Baltimore,” January 7, 2026: https://www.justice.gov/usao-md/pr/us-attorneys-office-leveraging-federal-state-local-partnerships-help-drive-sharp-decline
Baltimore City, 2025 Annual Report: https://s3.amazonaws.com/baltimorecity.gov.if-us-east-1/s3fs-public/2026-02/2025_annual_report.pdf
New York City Police Department, first-half 2026 crime statistics: https://www.nyc.gov/site/nypd/news/PR011/nypd-fewest-shooting-incidents-shooting-victims-murders-recorded-history-for
Sampson, Raudenbush, and Earls, “Neighborhoods and Violent Crime: A Multilevel Study of Collective Efficacy,” Science (1997): https://www.science.org/doi/10.1126/science.277.5328.918
Braga, Weisburd, and Turchan, “Focused Deterrence Strategies Effects on Crime,” Campbell Systematic Reviews: https://pmc.ncbi.nlm.nih.gov/articles/PMC8356499/
