The effect of ketorolac tromethamine, a non-steroidal anti-inflammatory drug, on postoperative blood loss and on the requirement for morphine was assessed after total knee arthroplasty, an operation in which blood loss is mainly measured rather than estimated. The purpose of this prospective, randomized, double-blind clinical trial was to determine whether administration of ketorolac in the perioperative period would increase bleeding related to the operation. Fifty-nine patients who had a total knee arthroplasty received either thirty milligrams of ketorolac or a placebo consisting of saline solution, intravenously, every six hours, in four doses. The first dose was administered about an hour before the end of the operation. Blood loss and use of morphine for pain control were measured for the first twenty-four hours postoperatively. The per cent change in the hematocrit and the amount of transfused blood were also recorded. The patients who received ketorolac used 27 per cent less morphine than those who received the placebo (40.0 +/- 23.4 milligrams compared with 55.1 +/- 23.5 milligrams [mean and standard deviation]); this difference was significant (p < 0.05). On the first day after the operation, the hematocrit decreased from 0.364 +/- 0.043 preoperatively to 0.278 +/- 0.032 in the patients who received ketorolac and from 0.363 +/- 0.046 to 0.298 +/- 0.030 in the patients who received the placebo. The 6 per cent greater decrease in the group that received ketorolac was significant (p < 0.05) but not clinically important.(ABSTRACT TRUNCATED AT 250 WORDS)