This study evaluated the yield and utility of urine toxicological screening (UTS) in psychiatric emergency room patients. We also assessed the effect of race, gender, and cocaine and/or alcohol detection on clinical symptoms and diagnoses. The study's null hypothesis is that recent use of cocaine and/or alcohol in psychiatric emergency patients does not change rates of psychiatric symptoms at evaluation. We did a retrospective chart review of 325 consecutive psychiatric emergency room patients and of 112 consecutive patients with benzoylecgonine (BE; major metabolite of cocaine) on UTS. At assessment clinicians were blind to the results of UTS in most cases. UTS was positive in about one-fourth of cases who submitted urine for toxicological testing. Alcohol, cocaine, and cannabis accounted for the majority of positive cases, and simultaneous detection of these drugs was common. Very few patients with BE on UTS were younger than 18 and none was older than 45. In patients with BE on UTS, substance use was missed in about one out of seven cases, and cocaine use in about one out of four cases. Patients with BE on UTS showed decreased rates of aggressive and bizarre behavior. Cocaine use increased rates of suicidality in males and rates of paranoia in black male patients. Black patients had more hallucinatory experiences than white patients, regardless of cocaine use. Principal diagnoses of psychotic disorders were frequently assigned to black patients with cocaine use, and principal diagnoses of affective disorders rarely. Patients with both BE and alcohol on UTS and patients with only BE on UTS had low rates of aggression, but patients with only alcohol on UTS were frequently aggressive. This study's findings show distinct clinical effects of cocaine and/or alcohol use on psychiatric emergency room patients, contrary to the null hypothesis. These effects differed according to race and gender. We conclude that there is little support to advocate routine toxicological testing in the psychiatric emergency room. Instead, selective testing based on empirical evaluation of risk factors might be more profitable to improve detection rates of cocaine and alcohol.