Objective: To determine whether older patients with chronic knee pain should be advised to use topical or oral non-steroidal anti-inflammatory drugs (NSAIDs). Design: Randomised controlled trial and parallel patient preference study. Setting: 26 general practices. Participants: People aged ≥50 with knee pain: 282 in randomised trial and 303 in preference study. Interventions: Advice to use topical or oral ibuprofen. Primary outcome measures: WOMAC (Western Ontario and McMaster Universities) osteoarthritis index, major and minor adverse effects. Results: Changes in global WOMAC scores at 12 months were equivalent. In the randomised trial the difference (topical minus oral) was two points (95% confidence interval -2 to 6); in the preference study, it was one point (-4 to 6). There were no differences in major adverse effects in the trial or study. The only significant differences in secondary outcomes were in the randomised trial. The oral group had more respiratory adverse effects (17% v 7%,95% confidence interval for difference -17% to -2%), the change in serum creatinine was 3.7 μmol/l less favourable (0.9 μmol/l to 6.5 μmol/l); and more participants changed treatments because of adverse effects (16% v 1%, -16% to -5%). In the topical group more participants had chronic pain grade III or IV at three months, and more participants changed treatment because of ineffectiveness. Conclusions: Advice to use oral or topical preparations has an equivalent effect on knee pain over one year, and there are more minor side effects with oral NSAIDs. Topical NSAIDs may be a useful alternative to oral NSAIDs.