OBJECTIVE: To identify the time point of the greatest degree of improvement in daily living activities, pain and depression in patients with osteoarthritis (OA) of the knee during 6 months of treatment with NSAIDs, in order to define compliance and drop-out rate.METHODS: 107 patients were recruited into a multicentre, prospective, randomized, controlled trial comparing two treatments, piroxicam-beta-cyclodextrin (PBCD) and slow release diclofenac (DCL).RESULTS: The greatest improvement in quality of life occurred in both groups after 3 months, with a slight further gain observed by the end of treatment. The Stanford Health Assessment Questionnaire score improved (p < 0.05 vs baseline) at 3 and 6 months with PBCD and at 6 months with DCL. The Arthritis Impact Measurement Scale score improved (p < 0.05 vs baseline) after 6 months in both groups. Significant (p < 0.05 vs baseline) improvement in other psychological and pain scores were recorded in both groups after 3 and 6 months. Compliance with treatment at 3 months was 73% for PBCD and 72% for DCL, and was 60% in both groups at 6 months.CONCLUSIONS: The results of this study indicate that the optimal length of time for an NSAID trial in OA patients is 3 months, when assessment of daily living activities is considered as the main outcome criterion.