The relationship between body mass index (BMI, Quetelet's index, kg m(-2)) and post-menopausal breast cancer risk was considered in age-specific strata on the basis of a pooled analysis of three Italian case-control studies, including a total of 3108 post-menopausal breast cancer patients aged 50 years or over and 2664 control subjects. Overall, there was a moderate, but significant, association between BMI and post-menopausal breast cancer: the odds ratios (ORs) were around 1.3 for the three intermediate quintiles compared with the lowest one, and 1.4 for the highest one. The association was moderate among women aged 50-59 years and 60-69 years, with ORs around 1.3 for the highest BMI quintiles, but stronger among elderly women, with ORs of 1.6 for the fourth and 2.1 for the fifth quintile. An 8-unit increase in BMI involved an OR of 1.18 at age 50-59 years, of 1.14 at age 60-69 years and of 1.59 above age 70 years. This pattern of risk is similar to that observed for post-menopausal hormone replacement treatment and is consistent with a duration-risk relationship in the exposure to high oestrogen levels and with a greater differential in oestrogen levels in overweight elderly women. In terms of population attributable risk, 19.6% of all post-menopausal breast cancer patients and 27.1% of those in women above age 70 years were attributable to overweight and obesity in this population. This has, therefore, major preventive implications as to reduce breast cancer risk late in life, it is essentially important to control weight gain in elderly women.