Bicycles are an alternative means of transport, promoted for health benefits. However, cycling is risky. In France the annual incidence of cycling injuries is estimated at 70 per 100,000. Most existing cycling safety studies are based on police data, which is inappropriate to cyclist safety research, being biased as to accident type (single vs. multi-vehicle crash) and recording as much as 10 times fewer cyclists than a road trauma registry for the same geographical area. The present study therefore used a postal survey sent to victims injured while cycling in 2009-2011 and identified in a road trauma registry. The survey sample comprised 1,078 injured cyclists. This article presents the construction of a Typology of Bicycle Crashes. A multifactorial approach is adopted with the Partitioning Around Medoids algorithm. Seventeen crash configurations compose the typology (7 utilitarian trips, 3 leisure rides and 7 sport practices), with 35 variables; the most discriminatory variables were "collide with another road user", "commutes", "practices sport" and "aged 60 and over". Each accident configuration is described by the proportion of its main characteristics in comparison with that in the overall sample. Some other factors, such as 'avoids another road user' (10%), "bad weather" (13%), "riding at night" (14%), "off-road ride" (20%) and "slips on the road" (40%), discriminated for collisions with another road user or obstacle. Injury severity in the 17 configurations is explored.