Interesting article. I did a little extrapolating on the numbers here. It states there were 51MM riders in 1985 and 40MM riders in 2004. Assuming the decrease in deaths was steady from 1981 to 2002 was constant we can see that deaths dropped at a rate of about 9.24/year. Using this we can estimate the deaths for 1985 at 924 and for 2004 at 748.5. This would give us a rate of 1.812 deaths per 100,000 riders in 1985 and 1.871 deaths per 100,000 riders in 2004. A slight increase in death rate. How much of that can be attributed to error and how meaningful that increase is open to debate, but it certainly seems that it has gotten any safer out there.