A previous retrospective study reported a 2-fold increase in hysterectomy rates in women outpatient clinical attenders with osteoarthritis (OA) compared to controls. Our study was undertaken to confirm the findings of a previous case-control study which suggested hysterectomy to be a risk factor for OA. A retrospective cohort study design was used that would reduce the problems of selection bias. One hundred and sixty-two women who had undergone a hysterectomy between 1978 and 1979 (current mean age 53.8) and 164 controls (mean age 54.1) were examined for peripheral joint OA between 1988 and 1989. The screening method was identical for both groups and included a questionnaire, systematic examination of certain joints and radiographs of hands and knees. Women with a previous hysterectomy were found to have significantly higher rates of clinical signs of knee OA and 1st carpometacarpal (CMC) OA than control women without hysterectomy. The results were confirmed when OA was classified by the presence of symptoms alone, and when only radiologically confirmed clinical cases were included. The application of radiological criteria showed significantly smaller medial joint spaces (by digital image analysis) in hysterectomized women, although no differences were found using the Kellgren and Lawrence grading system. The increased risk for knee and CMC persisted after adjustment for possible confounders including age, obesity, parity and smoking status. By contrast frequency of distal interphalangeal and proximal interphalangeal involvement was lower, though not significantly so, than in controls.