Retention in outpatient psychotherapy of 148 crack and cocaine abusers was examined. The clients were predominantly Black (63%) and Hispanic (21%), predominantly male (87%), and the majority (66%) had completed no more than 12 years of high school. Clients entered treatment in a low-cost treatment center in New York City between June 1987 and November 1988. Forty-two percent (62) of the subjects were seen for one or two research interviews only, and did not return to begin therapy. Of the 86 persons who came to at least one therapy session, 30% (26) dropped out before the third session, 28% (24) dropped out between the third and fifth sessions, and 42% (36) were retained for six or more sessions. Short-term and longer-term retentions were analyzed separately, using a battery including sociodemographic variables, treatment history, psychiatric symptomatology, number of arrests, and drug use variables. None of the variables considered was significantly related to short-term retention. There were large although not significant differences in longer term retention by therapist. Longer-term retention was associated significantly with being White (contrasted with being Black) and being young. Nonsignificant but large associations were found between longer-term retention and having few arrests, being Hispanic (contrasted with being Black), and having low SCL-90 scores. Results are compared with previous findings about retention in drug and alcohol treatment. It is suggested that future research on retention in treatment focus less on client variables and more on therapist and program variables.