With more than 23,000 affiliated vehicles, Uber accounts for 66 percent of all “black cars” in NYC. Crain’s recently reported that Uber nearly doubled its NYC fleet in the past year. Even as Uber adds several hundred cars a month, no one seems to be keeping track of how many traffic collisions involve Uber drivers, making it impossible to assess the company’s safety record.

Drivers affiliated with New York City Uber bases are known to have been involved in two fatal crashes in 2015. In January an Uber driver killed golf pro Wesley Mensing and injured his girlfriend Erin Sauchelli on the Upper East Side. The driver was reinstated by the TLC and Uber after a brief suspension. In May the driver of a car assigned to an Uber base killed 12-year-old Ervi Secundino in a school zone in Harlem. Witnesses said the driver was “flying” at the time of the collision.

After an Uber driver hit four kids and a mother on a sidewalk in the Bronx, Streetsblog asked around for Uber crash data to see how its safety record stacks up against other for-hire services. I checked with the Taxi and Limousine Commission, NYPD, and Uber. No one had an answer.

New York City has 35,528 active black cars — livery cabs reserved for pre-arranged trips by customers who generally pay with an account — and Uber bases account for 23,349 of them, according to the TLC. (Uber also has a luxury limousine base, with 657 affiliated vehicles.) All Uber drivers must have a TLC for-hire vehicle license, which is easier to obtain than a license to drive a yellow cab.

From July 2014 to June 2015, the latest period for which data is available, black cars were involved in 8,393 crashes, including four fatal collisions and two crashes resulting in critical injury, according to the TLC. Black cars are the only category of for-hire vehicle that saw a significant increase in that 12 month span, rising from 534 last June to 973 this July. While that change likely reflects the growth of Uber, it doesn’t tell us anything about the company’s relative safety.

Compounding the problem, the TLC collects mileage data only for yellow and green cabs. So there’s no way to calculate average crashes per mile for black cars and other types of for-hire vehicles based on data tracked by the city.

In addition to the different licensing requirements for black car drivers, Uber differs from medallion cabs in other ways that may affect how its drivers perform behind the wheel. The process of picking up customers is different, driver compensation is different, and Uber’s driver rating system has no analogy in the medallion industry.

The Uber model could be better or worse for public safety than the traditional one, or it may not be significantly different. With no one tracking the company’s crash data, there’s no way to tell.

We also asked app-hail company Lyft for crash data, and did not get a response. Lyft has just 533 affiliated vehicles in NYC, according to the TLC.

Uber and Lyft are notorious for keeping a tight lid on data about their trips. With app-hailing services poised to keep on growing in NYC, maybe the companies’ transparency about safety shouldn’t be optional.