Retrospective and prospective approaches to collecting information were compared in a study of spinal cord injury. The data show that the retrospective approach resulted in some underestimation of cases; that it was a very lengthy, time-consuming process; and that it produced significantly fewer useful clinical data than the prospective method.