For a given road and conditions, it's not how fast you drive, it's how much faster (or slower) you drive relative to the rest of traffic.
Of course, if you are on a road by yourself with no other traffic, then at a certain point too high of a speed will kill you because you'll run off of the road. But since most of us drive on roads with other traffic, it's the relative speed that is most important.
Below are some excerpts from various studies to support this basic fact (from a Dec 2 2000 posting in the Yahoo Groups forum for "Corbin Sparrow Electric Vehicle Driver's").
From "Speeding, Coordination, and the 55 MPH Limit."
Charles A. Lave
The American Economic Review, 75.5 (1985 December), pp. 1159 – 1163
"I find that there is no statistically discernible relationship between the fatality rate and average speed, though there is a strong relationship to speed variance…. Variance kills, not speed."
"Simple physics indicates that the consequence of a collision is a function of crash speed; and simple logic indicates that the probability of collision is a function of the dispersion of speeds on a given highway – more passing means more chances to collide."
"… all current safety campaigns emphasize that 'speed kills.' They imply that the slower driver is the virtuous one and is helping protect himself and other drivers. It isn't so. To reduce fatalities, it is important that everyone drive at about the same speed. Thus the major consideration in choosing a speed limit is that it be obeyed. And the major consideration for police is to reduce variation, not speed, because slow drivers are as much a public hazard as fast ones."
From "Driver Speed Behavior on U.S. Streets and Highways."
Samuel C. Tignor and Davey Warren.
Institute of Transportation Engineers: 1990 Compendium of Technical Papers, 1990 August, p. 85.
"The accident involvement rates on streets and highways in urban areas was highest for the slowest 5 percent of traffic, lowest for traffic in the 30 to 95 percentile range and increased for the fastest 5 percent of traffic."
From "Assessment of Current Speed Zoning Criteria."
David L. Harkey, H. Douglas Robertson, and Scott E. Davis.
Transportation Research Record, 1281 (1990), p. 51.
"In general, 85 percent compliance was achieved at speeds 10 mph [16 km/h] over the posted speed limit; … Speed at which accident risk is minimized occurred at the 90th percentile of the travel speeds observed."
From "Accidents, Overtaking and Speed Control."
Accident Analysis and Prevention, 3 (1971), p. 7.
"Driver education should bring to the driver's attention what his senses are incapable of telling him. Namely, that the risk to be involved in an accident is smallest in the vicinity of the median speed. 'Median speed driving' could be promoted also by advisory traffic signs."
"…a false notion of security induces some drivers into slow driving… the driver's intuition does not tell him that driving below the median speed increases his chances to be involved in accidents."
"The first question to be posed concerns the apparent irrationality of driving below the median speed. Seemingly, the slow driver accepts higher risks without any discernible reward (unlike the fast driver who is possibly trading safety for time savings). … the accident severity argument can hardly be used as a valid excuse for slow driving."