During a 3-year period, an attempt was made to develop a timely, representative, and sensitive firearm-related injury surveillance system. This system consisted of hospital and mortality records linked to police records of firearm incidents. Because lack of standardization of police department data precluded a statewide surveillance system, the focus was on the State's two largest urban areas, St. Louis and Kansas City. Firearm injuries occurring during crimes in the surveillance area in 1994 were recorded, and wounds inflicted unintentionally or during suicide attempts were excluded. An evaluation of the system showed it to be neither timely nor simple. Though estimated to represent 95 percent of the desired cases, information about firearms and circumstances of incidents was scant. It was determined that police records as they now exist cannot be included in a statewide firearms injury surveillance system and that the cost- benefit ratio does not justify even a regional surveillance system. Standardization of police records may be helpful but some information will always be lacking unless the perpetrator is arrested. 4 references and 2 tables