Crashworthiness of motorcycles has been evaluated over a period in excess of twenty—five years by means of crash testing motorcycles into automobiles. A common problem associated with this type of activity is the failure to maintain sufficient accuracy to allow repeatability. This is particularly true where both the automobile and the motorcycle are moving at the time of impact. This paper describes a highly accurate method of performing collisions between a moving motorcycle (with dummies aboard) and a moving automobile where the intent was to replicate actual on-road accidents. Summaries of eight demonstration collisions, which were performed between 1986 and 1995, are provided here. Each collision was predicated on an actual on~road accident. In each of the accident collisions the suggestion had been made that a steel loop type "crash bar" would have prevented injury. The object of these replicated collisions was to demonstrate the effects of these structures under actual accident conditions. These collisions were analyzed utilizing high speed 16mm film, real time videography and still photography. The photography, videography, crashed vehicles and dummies were reviewed to make measurements and comparisons between them and the actual persons and vehicles involved in the original accident.