Objectives To investigate changes in analgesic use before and after supervised exercise therapy and patient education in patients with knee or hip osteoarthritis (OA).Methods We recruited 16 499 of 25 933 eligible patients (64%; mean age 64.9; SD 9.6; 73% women) from the Good Life with osteoArthritis in Denmark (GLA:D) registry. Change in proportions of analgesic users (categorised according to analgesic risk profile; opioids > non-steroidal anti-inflammatory drugs > paracetamol) was assessed from before to after an 8-week supervised exercise therapy and patient education programme targeting knee or hip OA pain and functional limitations.Results Patients reported 13.2 mm (95% CI 12.8 to 13.6) less pain (visual analogue scale 0–100 mm) at follow-up compared with baseline. The proportion of analgesic users reduced from 62.2% (95% CI 61.5 to 63.0) at baseline to 44.1% (95% CI 43.3 to 44.9) at follow-up (absolute change: 18.1% (95% CI 17.3 to 19.0)). Among patients using analgesics at baseline, 52% changed to a lower risk analgesic or discontinued analgesic use. The proportion of opioid users after the exercise therapy was 2.5% (95% CI 2.1 to 2.9) lower than baseline; this represents a relative reduction of 36%.Conclusion Among patients with knee or hip OA using analgesics, more than half either discontinued analgesic use or shifted to lower risk analgesics following an 8-week structured exercise therapy and patient education programme (GLA:D). These data encourage randomised controlled trial evaluation of whether supervised exercise therapy, combined with patient education, can reduce analgesic use, including opioids, among patients with knee and hip OA pain.