Recent research has found that the police are involved with the majority of cases referred to urban mental health centers for involuntary civil commitment. In order to determine the basis for police decisionmaking, a random sample of 50 police officers were interviewed in Orange County, Calif., after they had petitioned a person for involuntary civil commitment; and a comparison random sample of 50 officers were interviewed after they had booked a person for arrest on a criminal charge. The officers had received a mean of 24 hours of special instruction in the handling of mentally disordered persons, and 94 percent had taken additional college courses in psychology or mental health. Their mean period of employment was 6.2 years. Ninety-five were male, and 86 were white (11 Mexican-Americans, two blacks, one Asian-American). The study found that (1) persons arrested were more likely to be male and nonwhite than those committed to a mental hospital, although within the arrested and committed groups, no significant sex or race differences were found in police perceptions of mental illness, dangerousness, or grave disabilities. (2) Virtually all persons petitioned for commitment were perceived as being mentally ill, while less than one-third of the arrested persons were considered mentally ill. (3) Committed persons were perceived to be slightly more likely to become violent to others, 5 times as likely to be gravely disabled, and 20 times as likely to be harmful to themselves. (4) In 30 percent of the arrest cases, the police could have petitioned for commitment, but they refrained because they did not believe that the degree of illness was sufficient to sustain a commitment. (5) In 30 percent of the commitment cases, the police could have effected a legal arrest. Police chose to commit rather than arrest people when they were faced with a choice. And (6) in virtually all of the commitment cases involving dangerousness to others, a physically assaultive act or threat had occurred within a day, and usually within an hour of the commitment. Yet in almost half of the commitment cases involving dangerousness to self, no attempted or overtly threatened act preceded commitment. These persons were committed either because they were viewed as incoherent or unable to provide food and housing for themselves, or because they were in a condition to be victimized by others. In sum, mentally disturbed persons were not being 'criminalized' by placement in jail as an alternative to commitment, and serious lawbreakers were not being placed in hospitals rather than in jails. The police did an accurate job of triage (treatment allocation) along the dimensions dictated by official public policy.