We report a prospective, randomised, controlled trial of the effect of either a non-steroidal anti-inflammatory drug (diclofenac sodium) or physiotherapy on the recovery of knee function after arthroscopy. At 42 days after surgery there was no significant benefit from either form of postoperative treatment compared with the control group. Complications attributable to the anti-inflammatory drug occurred in 9.6% of the patients so treated. Neither the routine administration of a non-steroidal anti-inflammatory agent nor routine physiotherapy is justified after arthroscopy of the knee.