The incidence of cancer was studied in a population-based cohort of 9,353 individuals (8,340 men and 1,013 women) with a discharge diagnosis of alcoholism in 1965-83, followed up for 19 years (mean 7.7). After exclusion of cancers in the first year of follow-up, 491 cancers were observed cf 343.2 expected through 1984 (standardized incidence ratio [SIR] = 1.4, 95 percent confidence interval [CI] = 1.3-1.6). A similar excess risk of cancer was seen among men (SIR = 1.4, CI = 1.3-1.6) and among women (SIR = 1.5, CI = 1.1-2.0). We observed the established associations with cancers of the oral cavity and pharynx (SIR = 4.1, CI = 2.9-5.7), esophagus (SIR = 6.8, CI = 4.5-9.9), larynx (SIR = 3.3, CI = 1.7-6.0), and lung (SIR = 2.1, CI = 1.7-2.6), although confounding by smoking likely increased these risk estimates. While there was evidence of increased risk for pancreatic cancer (SIR = 1.5, CI = 0.9-2.3), alcoholism did not elevate the incidence of cancer of the stomach (SIR = 0.9, CI = 6-1.4), large bowel (SIR = 1.1, CI = 0.8-1.5), prostate (SIR = 1.0, CI = 0.8-1.3), urinary bladder (SIR = 1.0, CI = 0.6-1.5), or of malignant melanoma (SIR = 0.9, CI = 0.3-1.9). Among women, the number of breast cancers observed was close to expected (SIR = 1.2, CI = 0.6-2.2), although a significant excess number of cervical cancers occurred (SIR = 4.2, CI = 1.5-9.1).(ABSTRACT TRUNCATED AT 250 WORDS)