Two nationwide surveys were carried out using an electronic poll of 2,000 GPs and postal questionnaires were sent to 30,000 patients with osteoarthritis (OA). Both surveys found a high level of gastro-intestinal (GI) side-effects in patients treated with non-steroidal anti-inflammatory drugs (NSAIDs). Almost every GP (97%) reported experience of patients suffering GI symptoms while on an NSAID, 38% reported patients who had been hospitalised and 4% reported patients who had died owing to NSAID-induced side-effects. Most GPs (92%) said they were concerned about GI safety when prescribing an NSAID and almost a third (32%) said they were concerned about litigation from patients who had experienced a bleed. Use of NSAIDs in OA remained high, with 44% of GPs prescribing conventional NSAIDs to at least three quarters of their patients, 57% of GPs using simple analgesia and just 12% using a cyclo-oxygenase-2 (COX-2) selective inhibitor in over 74% of patients. Some 45% of patients reported receiving NSAIDs compared with 43% on simple analgesia and 4% on COX-2 selective inhibitors. Most GPs (69%) stated that their main therapeutic objective in using an NSAID to treat OA was to control pain without GI side-effects. Almost a quarter (24%) said they used low-dose NSAIDs in the hope that this would control pain without GI side-effects. Dissatisfaction with treatment was the most common reason reported by GPs for patients on an NSAID to re-present, with 73% citing either breakthrough pain or incomplete pain relief as the most common reason for patient dissatisfaction. This mirrored the patients' perception, with 63% citing inadequate pain relief as their main reason for dissatisfaction with current painkillers compared to 17% who cited stomach upsets or irritation. Patient and GP appear to be united in their concern at the GI risks of NSAID treatment. In the light of this and recent data on the efficacy, safety profile and cost-effectiveness of COX-2 selective inhibitors, GPs should re-examine their medical management of OA.