OBJECTIVE: To investigate the influence of breast feeding, use of the oral contraceptive pill (OCP), and parity on rheumatoid arthritis (RA). METHODS: One hundred and seventy six women with RA were compared with 145 control subjects; all had at least one child. RA patients were classified as having severe (n = 82) or mild disease (n = 89) according to clinical joint evaluation, radiological score, biological inflammation, and the presence of HLA-DR1 or -DR4 alleles. RESULTS: The mean age of RA patients was 58 years, and the mean age at the time of diagnosis of RA was 46 years. The mean time between onset of RA and the first birth was 23.6 (SD 3.8) years. The OCP user rates were 33% in the RA group and 47.6% in the control group (p < 0.02). OCP use was related to the mother's year of birth. The relative risk for developing RA was 0.598 (95% confidence interval (CI) 0.33 to 1.1) in women who had used OCP for more than five years compared with those who had never used OCPs. In contrast, the age at which the first pregnancy occurred, the number of children breast fed, and the duration of breast feeding were comparable in RA patients and healthy subjects. Among the RA patients, parity, duration of breast feeding, and the number of breast fed children were significantly increased in those with severe disease. Having more than three children increased the risk of developing severe disease 4.8-fold when adjusted for age and OCP use. Forty six percent of women with severe RA had a history of breast feeding duration greater than six months before disease onset, compared with 26% of patients with mild disease (p < 0.008). Having more than three breast fed children increased the risk of poor disease prognosis 3.7-fold. In contrast, OCP use had a protective role in the course of RA (44% of RA patients with mild disease were OCP users, compared with 21.7% of those with severe RA; p < 0.001). Among those using OCP for more than five years, the relative risk of developing severe disease was 0.1 (95% CI 0.01 to 0.6), after adjustment for age, parity, and breast feeding. CONCLUSION: Our results suggest that parity, and to a lesser extent breast feeding, before RA onset worsened RA prognosis, whereas OCP use had a protective role. Prolactin and oestrogen may have a role in these effects.