Bicycle: .039 fatalities per million miles Car: .016 fatalities per million miles
http://www.bellboycott.com/cached/www.kenkifer.com/bikepages...
I think the biggest problem is the lack of control. If someone makes a mistake, has a seizure, loses control etc, and crashes into a cyclist, the cyclist has less 'protection' around them.
According to [0], 22 people were killed on bicycles in 2012. Compare that to 81 for an ordinary car, 31 pedestrians and 11 on a motorcycle.
Accoring to [1], bike rides make up 16% of all trips in Denmark, but only 3.6% of the actual number of kilometers travelled (which isn't really surprising). According to [2], car rides make up 46% of all trips.
So even though the number of car rides are only almost 3 times bigger than the number of bike rides, the number of fatal accidents are actually almost 4 times bigger.
Edit: Oh, passengers aren't included in the 46%. They make up 12% according to [2]. So my last point isn't completely valid, its more like they're equally fatal.
[1]: http://www.modelcenter.transport.dtu.dk/Transportvaneunderso...
[2]: http://www.modelcenter.transport.dtu.dk/Transportvaneunderso...
Edit: And it's not 3.6% of car kilometers, it's 3.6% of the total distance traveled, whether by car (as driver and passenger), bike, public transport, walking, motorbike and so on.
2nd edit: And for clarity, accoring to [2] above, car rides (driver + passenger) made up about 75% of kilometers traveled in 2012 in Denmark
Without some pretty sophisticated data, I'm not sure this is true. About 30,000 people die in car accidents every year, which I wrote more about here: http://blog.seliger.com/2013/11/10/foundations-give-away-fiv... . I don't know how many died on bikes or how many miles people bike versus how many they drive.
But car accidents are so common that even deaths often don't register on the news.
There's also an interesting question about the effects of biking and other exercise on health; riding may have salutary health benefits that driving doesn't. I don't know of any studies that examine such effects but would be happy to see any studies on the issue.