The basic methodology for the study involved compiling a database of suicide by police incidents, subjectively sorting these incidents into groups of similar incidents, defining the various categories thus created, and determining how reliably an independent judge could assign incidents to categories that had been created. Most police suicide incidents were obtained either from the professional literature or from newspaper databases. The 143 incidents consisted of three main categories: (1) direct confrontation in which suicidal subjects instigated attacks on police; (2) disturbed criminal interventions in which potentially suicidal subjects took advantage of police intervention; and (3) criminal interventions in which subjects preferred death to submission. When these three categories were sub-divided into nine types, two judges obtained a reliability coefficient of 0.87 for distinguishing suicide by police officer and 0.58 for placement into the nine types. Meaningful distinctions among types were found on three variables--subject age, real danger, and lethality. Implications of the findings are discussed, and more research is recommended to learn about suicides in police situations and how to handle them. 19 references, 2 notes, and 1 table