We don't generally waste a lot of time on this forum whining about how dangerous cycling is, because most of us know that for experienced and sensible cyclists the risk is not huge. Here's some confirmation, and enough information to let us get a statistical handle on what is meaningful and what isn't.
An American cyclist's chances of being injured while riding a bicycle in 1999-2003 has been measured as 1 in 68,400 journeys, and of being killed 1 in 4,761,904 journeys. (All other calculations are based on the same sample/time/reports cited below. Present tense because these are the best figures to be found, and likely to be directly applicable to our own period.)
A person's chances of being injured in a car are 1 in 124,533 journeys, and of being killed 1 in 10,869,565journeys.
Thus a car is twice as safe as a bicycle, or a bicycle is twice as dangerous as a car. Gee.
For perspective, let's say you commute to work five days a week and on both your days off take two journeys as well, 14 journeys a week. Then, if your fortune is precisely, randomly, average, you could ride for 93 years before suffering an injury.
Clearly, in such low risk activities as cycling or motoring, a differential risk factor of 2 isn't worth considering.
However, if the differential risk of being hurt on your bike was ten times that of getting hurt in a car, 124,533/10 or 1 in 12,453 journeys, then the average commuting/recreation rider would stand a chance of being hurt once every 17 years. Being hurt two or three times in a 50 year cycling career, while still a relatively low risk, is probably a consideration for most people..
So, for such comparatively very low risk activities as cycling, a difference of double the risk is just about negligible. What you need before the differential can be considered rationally is a differential of around a magnitude. Or higher.
Statistically, it isn't worth discussing the comparative risk of dying on your bike; nobody grows that old. However, there is a gathering perception in the cycling community that such statistical manipulations fail to meet the common perception that, statistically, none of us should know any cyclists who have been killed on a bike, yet in real life all of us know one or more or several cyclists dead on the roads.
***
Source material:
http://aje.oxfordjournals.org/content/166/2/212/T3.expansion.htmlhttp://aje.oxfordjournals.org/content/166/2/212/T3.expansion.html