In a cohort of 25,000 women aged 40–65 years at intake in a periodic screening programme for breast cancer the occurrence of endometrial cancer was studied during a follow-up period of up to 18 years. The authors examined whether they could confirm the existence of a number of relationships between this cancer and some reproductive and anthropometric factors which had been found in several case–control studies and in a few cohort studies. A comparison was made between 147 cases of endometrial cancer occurring during the period of follow-up and a random sample of 900 women taken from the cohort (334 being premenopausal and 566 postmenopausal on the day of intake). An inverse relationship between number of children and endometrial cancer risk was found in the older group; the excess risk among nulliparous women was stronger in married than in single women. Late age at menopause (after age 52) was associated with increased risk. Use of oestrogenic drugs on day of intake for alleviating perimenopausal complaints (reported in 8% of women) increased risk, especially in the younger group. Body weight was positively associated with risk (with an odds ratio of 4 in those over 80 kg) among postmenopausal women. Because there appeared to be a moderately strong risk associated with tall height, especially among postmenopausal women, Quetelet's index performed less well than body weight itself. Nevertheless, subscapular and triceps skinfold thicknesses confirmed an effect of fatness on risk. These results are discussed with reference to other epidemiological studies, in particular two cohort studies from Norway. The effect of height, convincingly shown by Tretli and Magnus in 1990, is explained in a way which does not assume causal mechanisms operating at a young age: absolute fat mass rather than relative weight is regarded as the main determinant of risk in postmenopausal women.