Injury accident data, pedestrian counts and vehicle flows have been studied for lengths of road on and near 140 pedestrian crossings. These crossings were located in similar conditions at sites throughout the country which were selected on the basis of having good visibility and not being close to busy junctions. Analysis of the data showed that daytime accident rates were closely related to the function P1/4V, where P is pedestrian flow and V is vehicle flow per hour, but that there was no consistent pattern between darkness accident rates and flows. There was no evidence of any differences in accident rates between Zebra crossings with and without refuges, nor of differences in pedestrian accident rates between Pelican and Zebra crossings. The analysis did provide evidence that Pelicans have a lower total injury accident rate than Zebras when the road length in the vicinity of the crossing is taken into account. It is estimated that, if operated under the same flow conditions, Pelicans would have approximately half an accident per year less than Zebras over the 100 yard length of road including the crossing