The short answer for New Yorkers heading to the train this week: serious crime on the subway system fell sharply in the most recent week of NYPD data, the 28-day rolling number is essentially flat compared to last year, and the year-to-date pace is tracking just below 2025 — which itself was the safest year on the subway since 2009 outside of the pandemic. The single most useful number is on the bottom-left of NYPD’s weekly CompStat sheet: 166 felony complaints in the subway system over the last 28 days, exactly the same as the same 28 days in 2025.
That’s what residents actually want to know — not the press release version, the version that helps you decide whether to take the train tonight. Below is the full decode of the numbers NYPD released for the week of May 4 through May 10, 2026, and what each line means if you’re the one swiping in at Atlantic, 14th Street, or Roosevelt Avenue.
What NYPD Actually Said This Week
According to the NYPD CompStat Citywide weekly report (Volume 33, Number 19, covering the week of 5/4/2026 through 5/10/2026), there were 37 major felony complaints inside the New York City subway system last week. That’s down from 43 the same week a year ago — a 14% week-over-week decline.
The “Transit” line on the CompStat sheet does not count every incident on a train. It counts the seven offenses NYPD tracks as the index crimes — murder, rape, robbery, felony assault, burglary, grand larceny, and grand larceny of a motor vehicle — that occurred inside the transit jurisdiction. Fare evasion, harassment, low-level disorderly conduct, and most subway behavior people complain about online are not in this number. That matters: when riders say “subway crime is up” or “subway crime is down,” they’re almost always reacting to the small bucket that includes the most serious incidents.
The Number That Actually Matters: The 28-Day Rolling Total
One week of data is noisy. A subway system that handles roughly four million weekday rides will see weekly numbers swing 10% to 20% just from random variation — a single robbery pattern in one station, a single big arrest at a transfer hub, holiday timing. The number to anchor on is the 28-day rolling figure.
For the four weeks ending May 10, 2026: 166 major felony complaints inside the subway system. For the same four weeks in 2025: 166. Change: 0.0%.
Translated for the rider deciding whether to take the F train home at 11 p.m.: across the entire 472-station subway system, the most serious crimes are occurring at the same rate as they did last spring — which was already the safest spring in 16 years outside the COVID-era drop in ridership. Average that 166 figure across 28 days and you get fewer than six felony incidents per day across the entire system. Not zero. But on a per-ride basis, statistically smaller than walking out of most non-subway public buildings in the city.
Year-to-Date: Tracking Just Under 2025’s Record Pace
Through May 10, 2026, NYPD’s CompStat shows 769 major felony complaints in the transit system, compared to 776 through the same date in 2025. That’s a 0.9% decline year-to-date. Over a two-year window, the system is running 0.6% below 2024. Over a sixteen-year window — the historical comparison NYPD prints right on the sheet — the transit number is 3.9% higher than the comparable period in 2010, when subway ridership was significantly lower.
For longer-horizon context, NYPD’s April 2, 2026 press release summarized Q1 2026 transit performance: 537 transit complaints in January through March compared to 544 in Q1 2025, a 1.3% decline. Within that Q1 figure, grand larceny in transit fell 9.1% (251 incidents vs. 276), which NYPD described as an all-time low excluding the 2021 pandemic period. Felony assault in the subway fell 6.6% in Q1 (155 vs. 166). Both numbers are notably better than the overall transit decline because most subway crime is theft and assault — when those two move, the whole transit number moves with them.
What That Means for the Crimes Residents Actually Worry About
“Grand larceny” sounds like a jargon term. In the subway, it almost always means one specific thing: stolen phones, wallets, AirPods, laptops, and bags — usually grabbed at the door as a train pulls out of a station, picked from a pocket on a crowded platform, or lifted from an open bag while the rider is looking at a screen. It’s the single most common serious crime category on the system and the one most rider-relevant for daily decision-making.
The Q1 2026 drop in that category — about 25 fewer victims compared to the same three months in 2025 — is the most directly useful data point for someone who wants to know whether the train is safer to ride this spring than last. The answer is: marginally yes, especially for theft, holding steady for everything else.
Felony assault is the second category most riders care about. That’s a serious physical attack causing injury, not a verbal incident or a shove. Down 6.6% in Q1 2026 versus Q1 2025. Robberies, burglaries, and homicides in the subway are extremely rare events — the Q1 2026 total of 537 across all seven categories works out to roughly six per day across the entire 24-hour, five-borough system.
How to Read These Numbers as a Daily Rider
Three practical takeaways from the CompStat sheet most New Yorkers won’t read directly:
One — the system is not getting worse. If your read of the subway is shaped by individual high-profile incidents that show up on local TV or social, the data does not support a story of an escalating crisis. The 28-day rolling number is flat. The year-to-date is slightly down. Both compare to a 2025 baseline that was itself unusually low historically.
Two — the system is also not improving as fast as the city overall. Citywide major crime is down 6.28% year-to-date on the same CompStat sheet. Transit is down 0.9%. That gap is real and worth naming. The subway is sustaining last year’s gains while the streets above are still trending down. NYPD attributes much of that gap to a deliberate strategy that pulled officers onto transit foot posts starting in late February.
Three — grand larceny is the crime you can most reduce yourself. Phone theft at the train doors and pickpocketing on crowded platforms account for the bulk of incidents. Keep the phone away from the open door as the train pulls out, zip the bag, watch the seat next to you. These aren’t rare-event protections — they’re the actual daily fact pattern behind almost every grand larceny on the system.
Where to Check the Same Numbers Yourself
Three primary sources behind the data in this article, each updated on its own schedule:
The NYPD Citywide Crime Statistics page hosts the weekly CompStat PDF and Excel file. New file every Monday morning covering the week ending the previous Sunday.
The CompStat 2.0 interactive portal lets riders search by transit district and pull line-level breakdowns that the weekly PDF aggregates.
The NYPD Transit and Bus Crime Reports archive posts monthly PDFs by transit district. Useful when a more granular question — say, what happened on the 2/5 line in Brooklyn this month — comes up. Monthly reports lag the weekly CompStat by several weeks.
The Bottom Line for Riders This Week
The subway in mid-May 2026 is statistically safer than the subway in mid-May of any year since the late 2000s outside of the pandemic-era ridership collapse. It is roughly as safe as the subway in mid-May 2025 — last year — and trending slightly better in the categories riders actually encounter, particularly theft. It is not improving as fast as the streets above ground, which is a real gap that the data shows clearly.
None of that means a particular train at a particular hour will be the system average. Late-night cars, certain transfer stations, and specific lines run hotter or colder than the system mean. But for the daily question of whether to take the train — the answer the CompStat sheet supports is the same one most regular riders already act on: yes, ride it, and zip your bag.
Data sources: NYPD CompStat Citywide weekly report covering the week of 5/4/2026 through 5/10/2026 (Volume 33, Number 19); NYPD press release dated April 2, 2026 reporting Q1 2026 statistics. Figures are preliminary and subject to revision.
