Data were collected on all psychiatric referrals to 10 emergency rooms in New York State during a 72-hour period in early December 1988. Overall, 30 percent of the 362 cases were brought by the police while the 10 emergency room percentages referrals varied from 10 to 53 percent. As compared with nonpolice referrals, police cases were as likely to be rated by clinicians as currently having psychotic symptoms, having a severe mental disorder, currently using substances, having a major mental illness diagnosis, and/or being referred to the emergency room for threatening to do harm or actually harming self. Police cases were more likely than nonpolice referrals to be male, to be referred to the emergency room due to impaired judgment or dangerous behavior to others, to be admitted to a psychiatric inpatient setting, and/or to have a longer length of stay in the emergency room. In addition, at municipal hospitals only, police referrals were more likely to be assaultive in the emergency room and/or not have payment resources. A logistic regression differentiated police cases from nonpolice cases on dangerous behavior to others and impaired judgment as reasons for referral, payment resources in municipal hospitals, and gender but not on any of the mental disorder variables. Policy implications for training and access to pretransport information and consultation are discussed.