Of 62 men with non-gonococcal urethritis who entered a study to assess compliance with treatment with oxytetracycline, only 33 could be evaluated. Traditional methods (interview and the absence of oxytetracycline in the urine) showed incomplete compliance in nine. Use of low dose phenobarbitone as a pharmacological marker showed incomplete compliance in a further five patients. In addition, phenobarbitone concentrations gave information on the extent to which individual patients had omitted treatment and provided direct, as opposed to circumstantial, evidence of good compliance by most (18) of those studied. Only three of the 33 patients whose compliance was assessed had evidence of continuing infection at follow up, and there was evidence of incomplete compliance in only one of these patients.