[mikeash]

I'd really like to see that broken down by type of transit, as well as deaths for transit passengers versus people killed in e.g. private cars involved in accidents with busses.

My guess, and this is very much a guess, is that buses will be disproportionately dangerous, and a fair number of those killed by busses are in cars that collide with them.

For one random example, the DC metro has seen 9 deaths in the past five years and appears to run at about 1.8 billion passenger-miles per year, for a fatality rate of around 0.1 deaths per 100 million passenger-miles (or 1 per billion). This is very rough as deaths are spiky (those nine deaths happened all at once almost five years ago) and the long-term average is tough to judge.

[DougWebb]

In my town and the towns around me (suburbs of NYC in New Jersey) there are a lot of deaths from people being hit by commuter trains. It's quite possible that we have more people dying that way than in traffic accidents, especially if you don't count deaths on the highways in the area which don't have any pedestrians.

The problem here is that the trains mostly run at ground level, and the stations are mostly completely open. It's very easy to walk around behind of or in front of the train while it's still in the station or approaching the station. We've had incidents where individuals or groups of commuters who just got off the train try to cross the track behind the train, assuming that the gates are down because the train hadn't left yet, only to be struck by a train arriving on the other track. There was a big public education campaign after the last incident like that, but the town had to resort to a police officer stationed there during rush hour handing out jay-walking tickets before people would stop trying to cross while the gates were down.

There's also a strong suspicion that a lot of these deaths are suicides, but that's pretty hard to prove.

[mikeash]

That's exactly the sort of thing I mean. I imagine that the safety of commuter trains with open stations and at-grade crossings must be vastly different from the safety of subways with closed stations and full grade separation. But I would love to see real numbers.