OBJECTIVE--To determine whether women with rheumatoid arthritis (RA) had differences in obstetric and gynaecological histories when compared with sisters without RA (controls) METHODS--Ninety eight RA discordant sister pairs, 36 of whom were identical for histocompatibility locus antigen (HLA-A, HLA-B, and HLA-Cw) types, were asked to recall their age of menarche, duration of use of contraceptive pill, pregnancy history, and age of menopause. RESULTS--The 98 siblings with RA had an older mean age of menarche (13.90 (95% confidence interval (95% CI) 13.56 to 14.24) years) than their sisters (13.49 (95% CI 13.22 to 13.76) years; mean difference within pairs 0.41, 95% CI 0.09 to 0.73 years, paired t test t = 2.54, p = 0.013). When the pairs were divided into identical HLA and non-identical HLA groups, the first showed no significant difference (mean difference 0.17 (95% CI -0.40 to 0.73) years), whereas the second did (mean difference 0.55 (95% CI 0.16 to 0.94) years, t = 2.80, p = 0.007). A multiple regression analysis to predict differences in menarche in the non-identical HLA sibling pairs failed to show any demographic or reproductive confounding variables. In 19 RA concordant sibling pairs, the seven HLA identical pairs had similar ages of menarche, whereas the 12 non-identical HLA pairs had interpair differences that narrowly missed significance (p = 0.054). All other obstetric and gynaecological variables were not significantly different within the pairs. CONCLUSIONS--The interpretations of these results are that either delayed menarche may predispose to or act as a marker of RA, or HLA linked genes are important in determining the age of menarche irrespective of disease state. This study fails to support a significant role for other obstetric and gynaecological variables in RA.