The relationship between body mass measures at diagnosis and/or at different ages and ovarian cancer risk was investigated using an Italian multicentre case-control study. The study, conducted between 1992 and 1999, included 1031 cases of incident, histologically-confirmed epithelial ovarian cancer and 2411 controls admitted to the same network of hospitals for acute non-neoplastic conditions. Odds ratios (OR) and 95% confidence intervals (CI) were obtained using unconditional multiple logistic regression analyses. Weight and body mass index (BMI, kg/m(2)) 1 year prior to diagnosis/interview were not associated with ovarian cancer risk. A direct association emerged with waist-to-hip ratio (W/H) (OR=1.45 in the highest category), particularly among women with stage I-II cancers. Cases also had a higher BMI at age 30 years (OR=1.22). Conversely, cases had lower weight gain between age 30 years and the year prior to diagnosis/interview, both for cases with stage I-II and those with stage III-IV cancers.