BACKGROUND: NSAIDs used for the treatment of osteoarthritis (OA) have dose-related risks for gastrointestinal, cardiovascular and renal adverse events (AEs), particularly in elderly patients. Topical NSAIDs reduce systemic NSAID exposure and may mitigate these risks. OBJECTIVE: To evaluate the safety and efficacy of topical diclofenac sodium 1% gel (DSG) versus vehicle in patients aged 25-64 or ≥65 years who have been diagnosed with knee OA. STUDY DESIGN: Pooled data from three 12-week, randomized, double-blind, parallel-group, multicentre trials. SETTING: US primary care, internal medicine, orthopaedic and rheumatology practices. PATIENTS: Aged ≥25 years with mild to moderate (Kellgren-Lawrence grade 1-3) knee OA. INTERVENTION: After a 1-week analgesic washout, patients applied 4 g of DSG or vehicle four times daily to one knee. Rescue paracetamol (acetaminophen) up to 4 g/day was allowed. MAIN OUTCOME MEASURE: Key efficacy outcomes common to the three trials were Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) pain (0-20) and physical function (0-68) subscales, global rating of disease (GRD; 100-mm visual analogue scale [VAS]) and pain on movement (POM; 100-mm VAS). ANOVA was used to compare efficacy outcome differences (DSG vs vehicle) by age (25-64 or ≥65 years). A flare design was used that defined a subset of patients who experienced increased pain during the washout period (modified efficacy subpopulation [MES]). RESULTS: The MES included both patients aged 25-64 (n = 602) and ≥65 (n = 374) years. Patients in each age group applied >90% of scheduled doses. Among patients aged 25-64 years, the improvement from baseline to week 12 (least squares mean [standard error]) was greater for DSG versus vehicle for WOMAC pain (-5.8 [0.3] vs -4.7 [0.3], p = 0.007), WOMAC physical function (-17.9 [0.9] vs -14.2 [0.9], p = 0.002), GRD (-29.5 [1.6] vs -23.8 [1.6], p = 0.01) and POM (-37.3 [1.8] vs -29.0 [1.8], p < 0.001). Among patients aged ≥65 years, the improvements from baseline for most efficacy outcome scores were significantly greater with DSG versus vehicle: WOMAC pain (-5.3 [0.3] vs -4.1 [0.4], p = 0.02), WOMAC physical function (-15.5 [1.1] vs -11.0 [1.1], p = 0.004) and POM (-33.7 [2.2] vs -26.4 [2.2], p = 0.02). The efficacy of DSG did not differ significantly between patients aged 25-64 years and ≥65 years: WOMAC pain (p = 0.85), WOMAC physical function (p = 0.70), GRD (p = 0.86) and POM (p = 0.81). The incidence of any AE was greater with DSG than with vehicle among patients aged 25-64 years (56.6% vs 50.8%) and ≥65 years (55.8% vs 43.9%). Treatment-related application site dermatitis was more common with DSG compared with vehicle in both younger (4.0% vs 0.7%, respectively) and older (5.8% vs 0.4%, respectively) patients and was the main reason for the difference in treatment-related AEs between the DSG and vehicle groups. Gastrointestinal AEs were infrequent among patients treated with DSG and similar to incidence rates with vehicle in both age groups. CONCLUSIONS: DSG was effective and generally well tolerated in adults regardless of age. These data support the topical application of DSG for relief of OA knee pain in elderly and younger patients. Clinicaltrials.gov registration numbers NCT00171626, NCT00171678, NCT00426621.