For a systematic review of postoperative analgesic efficacy and adverse effects of single doses, injected or oral, of pethidine and ketorolac compared with placebo, we sought published randomized studies in moderate to severe postoperative pain. Information on summed pain intensity or pain relief outcomes over 4-6 h was extracted and converted to dichotomous information to produce the number of patients with at least 50% pain relief. This was used to calculate the relative benefit and number-needed-to-treat (NNT) for one patient to achieve at least 50% pain relief. Minor and major adverse effect data were extracted and summarized. For pethidine 100 mg i.m., eight randomized, controlled studies met the inclusion criteria, with 203 patients given pethidine and 161 placebo. The NNT to produce at least 50% pain relief was 2.9 (95% confidence interval 2.3-3.9). At this dose, pethidine produced significantly more drowsiness and dizziness than placebo, with numbers-needed-to-harm (NNH) of 2.9 (2.2-4.4) and 7.2 (4.8-14), respectively. For ketorolac, 14 reports met the inclusion criteria (six i.m. and eight oral). Most i.m. information (176 patients) was available for the 30 mg dose, which had an NNT of 3.4 (2.5-4.9). Most oral information was available for the 10 mg dose, which had an NNT of 2.6 (2.3-3.1). Oral ketorolac 10 mg was consistently at least as effective as ketorolac 30 mg i.m. Only with oral ketorolac 10 mg were there significantly more adverse effects than with placebo, with an NNH for any adverse effect of 7.3 (4.7-17).