Background Best practice health and safety concentrates on removing, or reducing, sources of danger to achieve the largest reductions in individuals’ risk. However, analyses of personal travel have focused on the risks of different modes. The risks imposed on others by a given mode of travel have received little attention. Using data from two time-periods (2005–2007; 2008–2010) we examined changes in the comparative risk of third-party fatality by sex, age, and travel mode.

Methods The main outcome measures were the number of: (1) car driver and cyclist fatalities; and (2) third-party fatalities (e.g., passengers, occupants in other vehicles in multi-vehicle collisions, pedestrians, cyclists) by time spent travelling (fatalities per million hours’ use, f/mhu). Data on fatal car and bicycle accidents were extracted from STATS19, the national database of crashes on highways in Britain reported to and by the police. Variables recorded for each crash include details of all vehicles involved, age and gender of occupants, and resulting casualties. Data on time travelled by mode (exposure) was extracted from the National Travel Survey, and was aggregated to the population using mid-year population estimates. Subgroup differences in f/mhu were tested using Poisson regression.

Results Fatality rates for drivers (aged 17+) and cyclists (all ages) fell significantly from 0.093 f/mhu (95% CI 0.090–0.097) and 0.475 f/mhu (0.442–0.509) respectively in 2005–2007 to 0.067 f/mhu (0.064–0.070) and 0.323 f/mhu (0.294–0.353) in 2008–2010, reflecting decreases in the number of fatalities (both modes) and increases in usage (cycling). Age-specific rates showed U- and J-shapes for car drivers and cyclists respectively. Fatality rates for drivers aged 17–25 fell significantly from 0.372 f/mhu (0.353–0.392) in 2005–2007 to 0.252 f/mhu (0.234–0.271) in 2008–2010 (p < 0.001). Likewise, fatality rates for drivers aged 70+ decreased from 0.155 f/mhu (0.138–0.174) to 0.131 f/mhu (0.115–0.148) over the same period.

Third-party fatalities in car accidents occurred most frequently for young drivers, with rates for men more than double those for women. Third-party fatality rates decreased for male drivers aged 17–25 from 0.807 f/mhu (0.783–0.828) in 2005–2007 to 0.581 f/mhu (0.552–0.611) in 2008–2010, reflecting decreases in the number of non-driver deaths. Risks imposed on third-parties by car drivers were significantly (>30 times) greater than those by cyclists (0.157 (0.152–0.161) versus 0.024 (0.015–0.036) f/mhu in 2005–2007; 0.119 (0.115–0.123) versus 0.014 (0.008–0.023) f/mhu in 2008–2010).

Conclusion Official studies comparing risk across travel modes have ignored the wide variation in risk posed to other road users. Third-party fatality rates are considerably lower for crashes involving bicycles than cars.