All previous studies of intra-articular (IA) analgesic drugs for postarthroscopy pain have administered test-drugs at the end of the arthroscopic procedure, before any baseline pain could be assessed. Assay sensitivity has often not been documented or has been assumed to be present if a placebo control group had significant pain during the observation period. We present an improved study design employing an IA catheter for test-drug administration only in patients with moderate-to-severe baseline pain within 2h postoperatively. Using this technique we explored the incidence of moderate-to-severe pain and possible predisposing factors for pain through a close follow-up of all patients. The study incorporated an explanatory study of IA ketamine. A double-blind, double-dummy technique was used. Summed pain intensity differences 0-120 min after test medication was the primary outcome variable. Of 77 patients assessed for inclusion, only 45 had moderate or severe pain. Significantly more women (78%) than men (45%) had moderate-to-severe pain (P<0.005). Those not included continued to have no or mild pain and consumed less rescue analgesics than those who had high baseline pain. Mean baseline pain in the patient group with moderate or severe pain was 50mm on a 0-100 m visual analogue scale (VAS) (SD=15.1)(n=45). Mean VAS in the patient group with no or mild pain was 7.5mm (SD=8.7)(n=32). The new method for IA analgesic trials solves the problem with undesirable inclusion of patients with no or mild pain. We observed rapid onset and significant pain relief after IA injection of 10 ml saline with or without ketamine 10mg, but no difference between these two test medications. Intra-muscular ketamine 10mg showed significantly better early pain relief, global evaluation, and longer time to rescue analgesic, compared with IA ketamine 10mg.