Probably not the kind of stats you're looking for, but made interesting reading none the less..
Stats...
I'm not much of a statistician, but I get the impression that they suggest that mile for mile a cyclist is more likely to cause a fatal accident than a car driver!
Not sure where you get that from, the linked table shows injuries and deaths caused to pedestrians (only) over a period of 10 years ('98 - '07); it does not purport to show the numbers of fatal vehicular accidents, nor the number killed in vehicular accidents in those years. The figures are:
Pedestrians killed by collision with cyclist: 0 to 5 per year, average 2.9
Pedestrians seriously injured by collision with cyclist: 45 to 72 per year, average 55.5
Pedestrian injury numbers total (including above) from collision with cyclist: 225 to 346 per year, average 262.3
Pedestrians killed by collision with car: 433 - 673 per year, average 530.5
Pedestrians seriously injured by collision with car:5064 - 8063 per year, average 6298.8
Pedestrian injury numbers total (including above) from collision with car: 24169 - 37367 per year, average 30426.9
Now, to get the "mile for mile" statistics you'd need the actual numbers of miles traveled by cars and cyclists per year and these are not in the link (I'm not sure they are available, at least for cyclists). Then there is the bias that will result from the fact that an enormous proportion of car miles are on roads where pedestrians are not allowed (so that a "mile for mile" comparison is invalid on those grounds alone).
