Intra-articular injection of micro-crystalline corticosteroid is used to treat symptomatic osteoarthritis (OA) of the knee, but its duration of effect and efficacy are uncertain. From the observation that pain in OA of the knee can often be blocked by infiltration of the soft tissues at the patella margin with local anaesthetic, this study investigated an injection technique in which corticosteroid was infiltrated around the patella. Thirty-eight patients with radiologically demonstrable and painful OA of the knee were treated either with peripatellar or intra-articular methylprednisolone in a randomised double blind study. Assessments of response to either injection were made at one week, one month and three months. Eleven of 15 measures improved significantly over time in both groups, but the differences between groups were not statistically significant. Analysis of individual responses showed that the majority of good outcomes were in the peripatellar group. Five patients receiving intra-articular injections but only one receiving a peripatellar injection withdrew because of treatment failure. Peripatellar injection is an alternative method of local administration of corticosteroid which is highly effective in a proportion of patients and merits further evaluation.