Dioxin (2,3,7,8-tetrachlorodibenzo-para-dioxin, or TCDD) is a powerful carcinogen in experimental animals, whereas the evidence in humans is limited. We examined cancer mortality from 1976 to 1991 among residents of Seveso, Italy, which was highly contaminated after an industrial accident. The area was divided into zones with decreasing exposure to dioxin (A = highest, B = lower, R = lowest). The population of a surrounding noncontaminated area was used as a reference group. Zone A was small (11,516 person-years); in that zone, we saw a moderate increase in mortality from digestive cancer among women [relative risk (RR) = 1.5; 95% confidence interval (CI) = 0.5-3.5]. In zone B, we also saw excesses at digestive sites (83,610 person-years), 10 years after the accident. Women had an increased mortality from stomach cancer (RR = 2.4; 95% CI = 0.8-5.7), and men had increased mortality from rectal cancer (RR = 6.2; 95% CI = 1.7-15.9). Hematologic neoplasms were increased. The highest risks were seen in zone B for leukemia in men (RR = 3.1; 95% CI = 1.3-6.4), multiple myeloma in women (RR = 6.6; 95% CI = 1.8-16.8), and Hodgkin's disease in both genders (RR = 3.3; 95% CI = 0.4-11.9 in men; and RR = 6.5; 95% CI = 0.7-23.5 in women). Soft tissue sarcoma was elevated only among zone R males (256,408 person-years; RR = 2.1; 95% CI = 0.6-5.4). We found no increase for all-cancer mortality or major specific sites (for example, respiratory among males, breast among females). The specific excesses that we observed were not explained by bias or confounding, and their association with dioxin exposure is plausible. The follow-up is continuing.