In a randomised double-blind study 25 patients each were treated with amikacin (375 mg/d three times i.m.) or gentamicin (three times 80 mg/d i.m.). The dose was decreased according to a predetermined schedule if there was impaired renal function. Treatment indication was urinary tract infections of different severity with gram-negative organisms resistant to penicillin and cephalosporin. The bacteriological findings of a urine sample obtained by bladder puncture three and ten days after the end of treatment were the criterion of success. In addition, the number of white cells in the urine was taken into account. On the basis of various criteria, amikacin was found to give the higher success rate. The most important criterion, disappearance of the initially demonstrated microorganism, demonstrated a statistically significant advantage of amikacin over gentamicin.