The aim of this study was to examine associations between oral contraceptive (OC) use, body mass index (BMI = weight/height2) and prognosis in invasive breast cancer diagnosed before the age of 45. Survival analyses of a consecutive sample of breast cancer patients were undertaken. The cases were initially registered in a nationwide case-control study of OC use and risk of premenopausal breast cancer in Sweden and Norway. All 422 cases were under 45 years of age at diagnosis, and recruited from the reports to cancer registries (Sweden) or from surgical departments (Norway) during May 1984 through May 1985. Detailed information about OC exposure was obtained in the initial face-to-face interview. With Cox's proportional hazards analyses, a significantly lower hazard rate [relative hazard (RH) = 0.54; 0.31-0.94] was seen in short-term users (< 4 years)--but not in long-term (> or = 4 years) users--than in never-users of OC. Non-significant estimates for RHs lower than 1.0, i.e. better prognosis, with long recency (> 5 years) and latency (> or = 10 years) of OC use were noted. Prognosis was not influenced by age at first OC use or of its timing in relation to the first pregnancy. A higher BMI was associated with a poorer prognosis, RH 5.9 (2.0-17.8) for BMI > or = 29 versus BMI < 19, but BMI was not a confounder or an effect modifier of the association between OC use and prognosis. This study does not indicate that OC use prior to the diagnosis of breast cancer has any adverse effect on the prognosis, at least not in women under 45 years of age at diagnosis.