We report on the experience of five trauma receiving hospitals (four general hospitals and one spinal cord unit) in establishing a multicenter trauma registry (TR) for the purpose of data sharing. To ensure data comparability, all coders were oriented to standard data definitions and injury severity scaling. Coders and their physician sponsors met regularly to review data. Data presented for the four general hospitals from January through September 1992 address comparison of mortality rates, resource implications of isolated hip fractures, and the utility of knowing regional neurosurgical (NS) trauma volumes. Because of a statistically significant higher mortality rate at hospital 2, 7.2% versus 4.7% overall, mortality data were further characterized by patient age, mean ISS, and frequency of severe head injury. This still failed to explain the mortality difference. Hip fractures utilized 11,120 (26.3%) of the total 42,341 TR hospital days. Interhospital differences in median length of stay in this population suggest that greater resource efficiencies can be realized. Earlier questions about the value of including isolated hip fractures in the data set have been answered by understanding the resource implications of this population. Problems of NS coverage arising from a regional shortage of neurosurgeons can now be addressed with a better appreciation of the intraregional differences in NS volumes. Use of congruent data sets, combined with a collaborative approach, has stimulated the application of multicenter TR data to quality improvement, and utilization and regional planning issues.