Although higher relative weight is generally considered to increase the risk of breast cancer, several case-control studies have suggested that the reverse may be true among premenopausal women. The association between Quetelet's index (a measure of relative weight calculated as weight/height) and the subsequent incidence of breast cancer was therefore examined during four years of follow-up among a cohort of 121,964 US women who were 30-55 years of age in 1976. In contrast to women who had experienced natural menopause or bilateral oophorectomy, the incidence of breast cancer among premenopausal women decreased with higher levels of relative weight. Age-adjusted relative risks for increasing quintiles of Quetelet's index were 1.00, 0.90, 0.90, 0.73, and 0.66 (Mantel extension test for trend = -2.82, p = 0.005). This inverse association was not explained by known risk factors for breast cancer and was somewhat stronger when Quetelet's index was computed using reported weight at age 18 years. The excess incidence of breast cancer among lean premenopausal women, however, was limited to tumors that were less than 2.0 cm in diameter, were not associated with metastases to lymph nodes, and were well-differentiated. These findings suggest that the apparent excess risk of breast cancer among lean premenopausal women may result at least in part from easier, and thus earlier, diagnosis of less aggressive tumors.