Crime on the New York City Subway: How Rare Is it Really?

For nearly 120 years, the New York City subway system has served as a vital means of mass transportation for New Yorkers. Millions of workers, commuters, students, tourists and others rely on the subway as a safe, reliable and affordable means of transportation. But unlike competing transit options such as cabs and rideshares, riding the subway requires trust in a crowd of complete strangers sharing close quarters. 

With its vast geographic reach — over 665 miles of track and 472 subway stops — and the fact that it runs 24 hours a day every day of the year (unusual among U.S. cities), it is perhaps remarkable that the city’s subway system only saw 1,120 violent index crimes (the most serious crimes that police keep track of, including murder, rape, robbery, felony assault, burglary and grand larceny) last year, or a little under one for every million rides. For a different perspective, consider that around 2.5% of the city’s violent index crimes happened on the subway.

Subway crimes receive a lot of news coverage because they happen in a space that relatively few (though, with work from home, a growing number of) New Yorkers can afford to avoid. Many of us must rely on the subway to get around, and we are therefore a captive audience. When a crime happens in a dark alley late at night, that sort of situation is easily avoided. As the saying goes, nothing good happens after midnight. But when a person is pushed onto the subway tracks or assaulted on a train car during rush hour, it’s easy to imagine that such an attack could have happened to any one of us, or to a friend or loved one. 

The salience of subway attacks and the fear that they drive means that politicians are highly sensitive not only to the data but to the headlines that they invariably make. In response to concerns about safety on the city’s subways, Mayor Eric Adams recently announced that increased police presence on the subways is the “new normal,” potentially signaling his belief that re-establishing pre-pandemic norms would not happen organically. On March 6 of this year, Gov. Kathy Hochul announced that 750 National Guard soldiers would be deployed to boost safety on the subways, a move that is ordinarily reserved in response to pressing emergencies such as riots, natural disasters and medical emergencies. This measure, too, may well continue through the summer.

We are three researchers who have worked with city crime data for some time. One of us, in a past life, served as director of analytics at the NYPD. So it’s not uncommon for each of us to field questions about subway crime: What do we know about it? How common is it really? How unsafe are the subways compared to other public spaces? And what should we be doing about it? These are all important questions, but unfortunately we’ve only been able to supply frustratingly vague answers. Surprisingly, there has been hardly any empirical research on transit crime and, as a result, we know almost nothing about it other than through simple counts of subway crimes that journalists generate from time to time using New York City’s Open Data portal, or those that are put together in time for MTA board meetings.

What follows is the first in an installment of three pieces about crime on the nation’s largest subway system. We begin with a broad exploration of recent trends in the number of subway crimes, taking into consideration how crimes have changed on a per ride basis. We consider the nature of risk on the subway as it accumulates over time and is experienced by a typical rider. Our next piece will compare and contrast safety on the subway with safety in other public spaces in New York City, asking the critical — but unanswered — question: Are the subways actually more dangerous than other public spaces in the city? Finally, a third piece will use data to assess what policymakers can do to optimize safety on the city’s subway system. Throughout the pieces, we will call attention to the limitations of what we know and of the available data.

A tale of two narratives

As New Yorkers emerge from the hangover of the pandemic, a debate over subway safety has emerged, with two conflicting stories vying for dominance. One story amplified in the last year, with a focus on specific random attacks, especially people being pushed onto subway tracks, points to the fact that there has been a measurable rise in violent crime on the subway since 2019, concluding that the transit system has spiraled out of control and is no longer safe for New Yorkers. 

Others have instead emphasized that there are millions of subway rides each day, and very few report any serious crime. Last year, there were over 1.1 billion trips on the subway. That means there was one felony assault reported on the subway for every two million trips taken in the same year. For some, that small risk signals that riding the subway is safe enough to dispense with any concern.

Both narratives hold elements of truth: While the absolute number of violent crimes has risen since 2019, the relative risk per ride remains low. To the first point, there were 1,120 violent index crimes reported as occurring on the subway in 2023, compared to 935 in 2019, a 20% rise (Figure 1). Notably, though, the increase in crime is not confined to the city’s subways. Since 2019, reported violent index crime has increased more than 30% outside of the city’s subways, about 50% more than the 20% increase observed on the subway — partly a product of the fact that violent index crimes on the subway dipped in 2023, while violent incidents outside of the subway continued to rise.

Figure 1.

There’s another issue to consider. Ridership on the subway has fluctuated dramatically over the last few years. In 2019, almost 1.7 billion subway rides were recorded. In 2020, during the earliest phase of the COVID-19 pandemic, ridership plummeted by almost two-thirds, with only 640 million rides recorded. Ridership has since partially recovered, to over 1.1 billion rides in 2023. It’s hard to intuit how these changes in ridership balance against increases in reported violent crime. At a minimum, given that there were large fluctuations in ridership, raw counts figure to cast a misleading picture of the average rider experience.

The risk of violent index crimes more than doubled on a per ride basis in 2020, from roughly 1 in every 2 million rides to approximately 1 in every 750,000 rides. Those numbers are now in decline alongside the city’s emergence from the pandemic. However, crimes per ride remain about 40% higher than in 2019, a fact which may drive heightened fear even though the overall number of serious crimes has increased to a lesser degree (Figure 2). Notably, the rate of reported property crime per ride has not changed much in the last six years, underscoring that the public discourse is justified in focusing on changes in violent incidents on the subway.

One concern with per-ride calculations is that fare evasion increased during the pandemic. According to the MTA’s estimates, the share of subway riders who entered the system without paying the fare rose from 5-6% immediately before the pandemic to 12-14% during the last few years. Taking the official number of riders at face value therefore has the mechanical effect of pushing up the number of crimes per paying rider. To be more accurate in understanding per-rider crime rates, we incorporated these fare evasion estimates in the per-ride statistics reported here to account for riders who didn’t swipe in. 

Figure 2.

This per-ride perspective paints a very different picture than when considering overall counts. A reader who is looking at Figure 1 might conclude that violent crime on the subway has been headed in the wrong direction in the last few years. On the other hand, a reader who is looking at Figure 2 might conclude that these changes were initially triggered by societal disruptions in 2020, with the situation now improving. 

Regardless of one’s interpretation, for any individual ride on the subway, it’s extremely unlikely that a reported violent crime will occur. For the sake of comparison, consider the average risk of injury or death from a car accident in the U.S. In 2022, American drivers faced about a one-in-one-million risk of injury or death for every mile driven — roughly equal to the violent index crime risk from a single ride on the subway in 2023. And yet one is hard-pressed to find news articles about Americans who are hesitant to drive one mile for fear of injury or death. So why has there been such a focus on the subway system? 

One potential reason is that reported index crime rates don’t capture the whole picture. Of course, the data we analyze here don’t capture incidents that were never reported to public safety officials in the first place. But also, for an assault to be classified as a felony-level index crime, it must be considered “serious” — e.g., so severe that it causes permanent physical damage, or the incident involves a weapon of some sort. Reported incidents that don’t meet this high bar are often recorded as misdemeanor assaults or even harassment, which is considered a simple violation, a level below misdemeanors. These less serious incidents are rarely discussed by city officials or covered in the press in favor of the most serious incidents.

But it turns out that lower-level violent crimes on the subway occur more frequently than felony-level assaults and robberies (Figure 3). As a result, these lower-level crimes are more likely to inform the public’s experience with crime on the subway. By our count, there were about three times as many reported lower-level violent crimes compared to reported felony-level violent crimes in 2023:

Figure 3.

Even after expanding the definition of what one might consider a “violent” crime to include misdemeanors and violations, there was only one reported violent incident for every 280,000 rides. Looking at this per ride rate in isolation could reasonably lead to the conclusion that fears of riding the subway are overblown.

What per-ride counts don’t capture

Here’s the wrinkle: While it’s very unlikely that any individual ride will result in the rider being the victim of a reported crime, over the course of a year, it becomes much more likely that you or someone in your community might see some such event on the subway, a feature of public transit which has been noted by others. 

For the sake of illustration, let’s start with the 2023 rate of one reported violent incident for every 280,000 rides. Then imagine that you take the subway about 10 times a week, resulting in 500 rides per year. Also assume that you can see 20 riders whenever you take the train – if anything, a conservative assumption. Finally, let’s say you keep in regular contact with about 30 people who live in New York and have the same subway ridership habits. 

We’ll save you from having to dig out your probability textbook by calculating some of the relevant risks for you. You still have a low probability — about 1-in-500 — of being a victim of a reported violent crime yourself, even over the course of an entire year. But probabilities have a surprising way of accumulating. You have a 1-in-30 chance of seeing a violent crime unfolding on the subway over the course of the year. And across your community of 30 people, there’s about a two-thirds chance that someone you know will have seen or experienced a violent crime unfolding on the train. New Yorkers who live in neighborhoods with higher crime, who take the subway during off-peak hours or who have longer commutes, likely face higher risks than this citywide average. For these riders – who also figure to be among the city’s less affluent residents – hearing first-hand accounts of subway crime may, in fact, be quite common. 

These probabilities are sensitive to the assumptions we outlined above, so they should be taken with a big grain of salt. But, in our view, the calculation is nevertheless instructive. Regardless of the precise risk that subway riders face, it’s clear that while most people may be at low risk of experiencing subway crime themselves, particularly if they’re occasional subway riders, there’s a much higher chance they will have seen or heard of something concerning from their network over the last few years. This discrepancy may explain why people may have grown concerned about subway crime alongside increased political and media attention, and clarifies how the competing narratives in today’s public discourse can both be touching on deeper truths about people’s experience on the subway. Beyond the numbers, it bears mentioning that even when the probability of violent crime is low, the persistent presence of people who are apparently suffering from mental illness and the general sense of chaos on the subways can undermine riders’ sense of security. 

The public discussion about safety on the subway has recently acquired a great deal of political salience for those in charge. That could be explained by the calculation above, which suggests that the risk of subway crime may have risen enough to be noticed by a typical rider. But it’s also possible that our newfound concern for safety on the subway is a vestige of many of us avoiding the subway during the pandemic, and reassessing what it means to go back underground and share close quarters with strangers. As Brandon del Pozo noted a couple years ago, “riding the subway is the paradigmatic act of social cooperation in a city’s public spa​​ce.” It’s possible that some New Yorkers have grown more fearful in the interluding years, having forgotten what it’s like to live amidst the persistent need for social cooperation.

We hope that this analysis helps to provide some much-needed perspective about the frequency of subway crimes. But it also prompts some additional questions. If we are concerned about safety on the subway, should we not be equally concerned about safety in other public spaces? How different is it to ride the subway compared to taking a walk down the street? Are we particularly unsafe when we ride the subway, or is it a space of relative safety — of heightened public trust — compared to other public spaces? And what can public officials do about it, anyway? These are worthy questions to consider, and something we’ll be investigating in the coming months. In the meantime, we won’t stop running down the stairs trying to catch the next train.