One-hundred and ninety-five consecutive psychiatric outpatients had urine sampled for the presence of opiates, cocaine, barbiturates, and amphetamines. All patients had demographic variables, diagnostic information, and treatment course analyzed in a blind fashion using preselected criteria. Patterns of abuse were defined for the 13.3% of the total sample who were covert abusers. An analysis of diagnostic accuracy for abusing and nonabusing populations indicated that covert abuse markedly distorted diagnosis and management. The study indicates that deme graphic variables and previous history were not reliable predicators of covert abuse. Drug abuse was a major factor in distorting the accuracy of diagnosis. Drug abusers were significantly less likely to improve with therapy than were nonabusers. Barbiturate abusers were most likely to suffer from adverse drug reactions. Urine screening for drugs was the single most valuable diagnostic tool and was felt to be indicated in a 11 cases where diagnostic confusion was present. Therapists saw the covert abuser as sick out of proportion to their social behavior, were perplexed and frightened by these patients' dependency, and perceived them as too brittle for therapy. Therapists behaved very differently with these patients than with diagnostically matched controls. They misdiagnosed them four times more often, missed appointments with them seven times more often, and were ten times more likely to refer them to another therapist or agency. Therapist anxiety and "flight from therapy" with covertly abusing patients are discussed.