Cancer risk following treatment with non-contraceptive estrogens was studied in a population-based cohort of 23,244 women. Complete follow-up for an average of 6.7 years revealed 1,087 incident cancers versus 962.5 expected (relative risk/RR/ = 1.13; 95% confidence interval 1.10-1.20). We confirmed the recent findings of a more detailed analysis of the same cohort, based on a 1-year shorter follow-up period, namely: a markedly increased risk of endometrial cancer (RR = 1.8; 1.5-2.1), notably in women receiving potent estrogens, i.e., conjugated estrogens or estradiol (RR = 2.0; 1.6-2.4), and a slightly increased risk of breast cancer (RR = 1.1; 1.0-1.2). A slightly decreased risk of invasive cervical cancer (RR = 0.8; 0.5-1.2) is most likely due to more frequent smear taking than in the background population. There was no increase in the risk of cancer of ovary (RR = 1.0; 0.8-1.2), pancreas (RR = 0.8; 0.5-1.2), large bowel (RR = 1.0; 0.8-1.2) or kidney (RR = 1.0; 0.7-1.4). The risk of developing cancer in liver or biliary tract was lower than expected (RR = 0.4; 0.2-0.7), particularly in women who had used potent estrogens (RR = 0.3; 0.1-0.6), an unexpected finding which warrants further studies. Increased risks of malignant melanoma (RR = 1.5; 1.0-2.1) and lung cancer (RR = 1.3; 0.9-1.7) need cautious interpretation because of their low magnitude, the absence of a biological gradient when subgroups were analyzed and the slightly higher prevalence of smokers in the cohort than in the background population.