Version 1.04, May 23, 2009 - Found NHSA 2003 study, DOT HS 809 585, showing absolute rates of pedestrian accident manuvers.
Version 1.03, December 21, 2009 - Covered "Straight line" incidents and SUV/pickup risks.
Version 1.02, November 10, 2009 - Added Table 6a quote showing the math problem.
Version 1.01, October 31, 2009 - Posted more conservative Prius fatality rate based upon per accident accounting Prius contribution.
Version 1.00, October 17, 2009 -

SUMMARY

The Prius fatality rate of 0.64 per 100 million miles is less than half of the NHTSA 1.37-1.51 fatalities per 100 million miles for the years 2001-2007. This is based upon the annual Prius sales from 2000-2007 minus an overly high estimate of 3% Prius lost per year; an annual usage of 15,000 miles per Prius per year; and 75 prorated deaths of the 124 fatalities from 111 fatal accidents including Prius and non-Prius vehicles. A more conservative accounting of Prius contributions, a count of Prius participation by accident, gives an even lower rate, 0.55 Prius fatalities per 100 million miles.

Details

The Prius went on sale in the USA in 2000 and by the end of 2007, 515,944 had been sold. However, only half of 2007 vehicles would have seen a full 15,000 miles of service bringing the effective fleet count to 405 thousand at the end of 2001-2007. Furthermore, some of these vehicles were lost by accidents, theft, and other disasters. Using the number of vehicles retired in 2005, 6.5 m. ("Copper in End-of-Life Vehicle Recycling", Emillio Brahmst, Center for Automotive Research, November 2006, pp. 6,) and 242.7 m. vehicles ("Traffic Safety Facts 2005 Data", Table 2, NHTSA, DOT HS 810 623,) we rounded up to an overly high estimate of 3% Prius loss per year even though the average age of Prius was much lower than the fleet average.

We used 15,000 miles per year per vehicle, the same number the EPA uses for annual mileage, to calculate the Prius miles per year, 2001-2007, and came up with 11.8 billion miles.

Using the National Highway Traffic Safety Administration (NHTSA) Fatality Analysis Reporting System (FARS), we found 111 accident records that included a Prius with a total of 124 fatalities. However, the Prius was less than 60% of the vehicles involved in these accidents giving a pro-rated, Prius fatality count of 75 deaths 2001-2007. Divide these 75 deaths by the 11.8 billion miles and scale it to deaths per 100 million miles and the number is 0.64 per 100 million miles for 2001-2007.

During this same interval, the NHTSA reported 1.51 to 1.37 fatalities per 100 million miles, more than twice the fatality rate of the Prius. There have been 11 Prius-only, pedestrian fatalities in this same period. This gives a Prius-only pedestrian fatality rate of .093 pedestrian fatalities per 100 million miles.

Skeptics

Recently, the NHTSA released "Incidence of Pedestrian and Bicyclists Crashes by Hybrid Electric Passenger Vehicles" (DOT HS 811 204, September 2009) that reports the relative distribution of pedestrian involved accidents without calculating the rate of accidents per hybrid vehicle or vehicle-mile. The Prius has an absolute fatality rate half that of ordinary vehicles so it is hard to see where the Prius would be more hazardous, factually "deadly," than the NHTSA published, a known hazard of all USA vehicles. Many readers of the report miss the semantic trick of "incident rate," which means within the self-selected accidents. Without a rate per vehicle or vehicle-mile, DOT HS 811 204, remains flawed and easily misinterpreted. For example, Table 6a, pp. 13 from the report:

The first clue was the ratio of incidents to percentages made no sense. The percentages were not per vehicle but something else that is not proportional to the risk by vehicle or vehicle mile. It turns out these are ratios based upon types of maneuvers, a profile, not an absolute risk per vehicle.

Only turning and backing up have any different profiles between the HEV and ICE vehicles and these constitute 13% and 2% of each vehicle groups' total accident involvement (Table 3d, pp. 11.) But this reveals another problem, a trick percentages can play on absolute incident counts:

Althought the percentage of "Going Straight Incidents" varies only by 3% the count for that 3% would be 109 incidents, which is over three times the count of "Turning" and "Backing" incidents.

Another problem is only similarly sized hybrid and ICE vehicles were compared, which is fair if trying to find out if the hybrid power system had a different risk compared to a compatible sized ICE power system. But the relative risks of SUVs and pickups, "the Elephant in the Room," were ignored. In Docket No. NHTSA-2009-0041, the NHTSA reported:

We can only imagine the results if both HEV and ICE vehicles were compared to SUV and pickups.

This report is flawed by small numbers of turning and backing incidents; failure to account for differences in "Going straight;" and ignoring SUVs and ICE vehicles. However, this NHTSA report gives a clue where HEV vehicle relative distribution of pedestrian accidents might give a small improvement in safety ... a very small improvement. It does not report the relative risk of HEV versus ICE accidents. Regardless of the types of pedestrian accidents, the Prius fatality rate remains half of all vehicle fatalities 2001-07.

In April 2003, the NHTSA conducted "Analysis of Pedestrian Crashes," DOT HS 809 585, and found:

Here is the single largest contribution to pedestrian accidents, and the one that the Prius has the lowest relative rate. Yet, it is the one ignored in the latest report, DOT HS 811 284.

"Quality Planning" reported in July that hybrid drivers average 25% more "non-commuting miles" than ordinary drivers or roughly an extra 2,000 miles per year. People who have to drive further distances are going to be attracted to fuel efficient hybrids. Quality Planning went on to report a 65% higher rate of traffic citations as a number likely reflecting driving in a denser, urban area with more miles than an average driver. Yet in spite of these reported, additional risk factors, the Prius remains less deadly by half than ordinary vehicles.

CONCLUSIONS

We know the Prius has half the fatality rate per 100 million miles as the NHTSA has reported for years 2001-07. But neither the NHTSA nor "Quality Planning" have yet to convert any of their data into a risk per vehicle mile, the Gold Standard the NHTSA uses. "Quality Planning" asserts there is a higher per vehicle risk without reporting the risks of non-hybrids with the same usage pattern.