The value of the gram-stained urethral smear in clinical decision-making was assessed in a study of 250 men attending a clinic for sexually transmitted diseases. Of the 250 men, 132 (52.8%) had objective evidence of urethritis. Neisseria gonorrhoeae and/or Chlamydia trachomatis was isolated from 94 patients (37.6%). No pathogens were isolated from 38 patients (15.2%) who were diagnosed as having urethritis. Although the specificity (0.95) and positive predictive value (0.95) of the gram smear for culture-proved urethral infection was high, the relatively low sensitivity (0.66) and negative predictive value (0.63), led us to conclude that the test was of limited value in diagnosis and therapeutic decision-making when the patient was first seen. The decision to treat a patient should be based on a reliable history of dysuria and/or a urethral discharge in a patient at risk of infection, with or without an observable urethral discharge. Nevertheless, a gram smear should be done for all patients who are diagnosed presumptively as having urethritis, because it may be the only objective evidence of urethritis.