Total homicides (i.e. ANYWHERE in the US, not specifically on a subway which is probably a tiny fraction below 1%) are roughly double the total deaths in motorised traffic accident. So I'd say it's a safe bet to take the subway over a car.
There are other threats of course, pickpocketing etc... but threats to my life specifically on a subway would be a total non-concern for me.
> You mean, the total homicides (the violent crime ones) are about half the rate of motor vehicle deaths. Actually closer to a third.
Whoops yes I accidentally flipped it around incorrectly weakening my own argument :p
> It's about 10-15k homicides v 40-50k car crash deaths per annum in the past decade.
I'd still say it's closer to double though, there haven't been as few as 10k murders since 1965, and as the population was much smaller then (about a third smaller) the rate of homicide per capita wasn't much different from the average of the past decade. Meanwhile car deaths haven't been 50k since 1980 so 10-15k and 40-50k would include numbers going back 3 to 5 decades.
For 2005-2015 though the motor vehicle deaths average about 11 per 100k[0], for murder it's about 5 per 100k.[1]
[1] https://en.wikipedia.org/wiki/List_of_motor_vehicle_deaths_i... (doesn't include 2014 and 2015 but they're estimated at roughly the 2013 numbers. Notice the substantial dip right around the recession, I wonder how much of the drop can be explained by fewer miles driven rather than improved safety).
There are other threats of course, pickpocketing etc... but threats to my life specifically on a subway would be a total non-concern for me.