Additional sources of information were questionnaires completed by patrol officers regarding their opinions on social service agencies and interviews with 103 agencies that accepted police referrals. About 40 percent of the observed referrals were direct in that police took citizens to helping agencies or internal police units or contacted the agency on the individual's behalf. The remainder were suggested referrals where the officer gave the citizen a name, address, or phone number. Referrals were observed in 13 percent of police-citizen encounters and involved 11 percent of their participants. When only direct referrals were considered, the rate dropped by more than half. No clear-cut determinant of referral rates emerged. Characteristics and attitudes of citizens had almost no bearing on the likelihood that a referral would occur. Victims and prior complainants were more likely than other participants to be referred. Referral was most common in violent crimes because victims often needed medical assistance. Police also referred citizens who had information they wanted and functioned as information brokers by calling a doctor or a towing truck. Officer age had no significant impact on referral, although analysis suggested that longer service increased the chances of referral. Officers' perceptions of referral agency availability had little effect on their behavior. The police complained that referral agencies were not open on weekends and evenings when they needed them most, while many agencies felt their sevices were underutilized by the police. Tables, footnotes, and 17 references are included.