There's an intriguing graph in the latest issue of A to B, in which the number of fatalities per billion cycle/kilometers (effectively, the individual risk of death per cyclist, corrected for distance) is plotted against the average cycle distance per person per day. Fatalities increase very rapidly with decreasing cycle distance, with the UK falling somewhere near the bottom of the curve, with an average cycle distance per person of about 0.25 km/day. What this means is that as people cycle less, the risk of fatality for the remaining stalwarts increases sharply.
The theory is that this is due to drivers becoming more vigilant and careful about cyclists when they are common, which seems like a plausible explanation. In fact, there is apparently a power law (proposed by a Mr. Sneed in 1949) that suggests that the risk of collision with cars is as follows1:
where n is the number of cyclists or pedestrians.
The graph was in the context of the possible introduction of compulsory cycle helmets — currently a very contentious issue for cyclists. The point is that in all countries where cycle helmets have been made compulsory, the number of cyclists has decreased. So on this evidence, the lower total number of fatalities might come with an increased risk of fatality to individuals. Given that many people are dubious about whether helmet use decreases or increases the severity of injury anyway, and that decreasing cycle use is likely to increase the burden on the NHS through increased obesity and cardiac problems, compelling people to wear helmets starts to look like a very bad idea. I do wear a helmet myself, but I'm not in favour of making everybody wear them.
1 You can learn things here, you know...