A total of 243 patients were entered at 20 general practice centres in the UK in this 6 week, double-blind, randomised, parallel-group study, which was designed to compare the safety and efficacy of nabumetone (1g at night) and slow-release diclofenac (100mg at night) in the treatment of osteoarthritis.Seven parameters of efficacy were measured, adverse events were recorded and laboratory monitoring was carried out.At the end of the study, both treatment groups showed highly significant clinical improvement with no significant intergroup differences.24 patients (20%) in the nabumetone group and 21 (17%) in the slow-release diclofenac group withdrew before the end of 6 weeks, mainly because of adverse events. Although overall the incidences of adverse events were similar in both treatment groups, significantly fewer patients complained of digestive system side effects in the nabumetone group (21) than in the slow-release diclofenac group (35) [p = 0.03].The study was also intended to evaluate the effectiveness of study monitoring by a clinical research associate who attended study centres with a laptop microcomputer to enter trial data. Immediate correction by the trialist of errors and omissions in the data, with on-site data entry on the laptop computer, contributed to trialist motivation and resulted in rapid recruitment of patients, collection of high quality data and speedy analysis at the end of the trial.