The influence of sex hormone related events and smoking on the development of OA in women was investigated in a case-controlled postal survey. One hundred and twenty-nine patients with nodal generalized osteoarthritis (NGOA) and 145 with non-nodal pauciarticular large joint osteoarthritis (LJOA) were identified from the database of a Nottingham OA clinic. For each patient, three age-matched controls were randomly selected from the same general practice. Sixty-three per cent of questionnaires (690/1096) were returned: NGOA, 95; NGOA controls, 226; LJOA, 113; LJOA controls, 256. There were no differences in age at menarche or menopause, rates of hysterectomy, oral contraceptive use, or hormone replacement use between cases and controls. Fewer OA patients had ever smoked [(Odds Ratio (OR) 0.65, 95% Confidence Interval (CI) 0.45-0.95)] and subset analysis demonstrated that this negative association occurred only in the LJOA group (OR 0.43, CI 0.25-0.72), particularly in those with knee OA (OR 0.29, CI 0.14-0.62). A previous successful pregnancy was negatively associated with NGOA (OR 0.47, CI 0.24-0.95). This study demonstrates no association between oestrogen-related hormonal events and OA, but a negative association between smoking and LJOA. Such data supports the concept that OA is a heterogeneous disease and underlines the need to differentiate OA subsets.