Thirty-seven patients with chronic back pain were entered into a randomised, 3-way, double-blind, cross-over comparison of naproxen sodium 550 mg twice daily, diflunisal 500 mg twice daily, and placebo. Each treatment was given for 14 days after a preadmission wash-out week during which only paracetamol was allowed. Patients were assessed on admission and at the end of each treatment with respect to global pain, night pain, pain on movement, and pain on standing. Both visual analogue scales and simple descriptive scales were used to measure pain. Side effects were elicited by a nonleading question. Both methods of pain measurement gave similar results and were highly correlated. Naproxen sodium was superior to placebo in relieving global pain and depending on the method of measurement, in relieving night pain and pain on movement. Diflunisal showed no significant differences from placebo. Side effects were similar on all 3 treatments. The final preference of the patients was significantly in favour of the active treatments.