A case-control study of uterine cervical cancer was conducted using 331 cases and 993 age-matched controls identified through the Missouri Cancer Registry during 1984-1986. Patients with smoking- or alcohol-related cancers were excluded from the control series. Logistic regression was used to compute odds ratios (ORs) and 95% confidence intervals (Cls) after adjustment for age, cigarette smoking, alcohol consumption, and stage at diagnosis. A dose-response relation was observed between intensity of cigarette smoking and invasive cervical cancer, with light and heavy smokers having elevated risks (OR = 2.2, 95% Cl = 1.4-3.6 and OR = 3.9, 95% Cl = 2.7-5.6, respectively). Former smokers had less elevated risk (OR = 1.7, 95% Cl = 1.0-2.9), a finding consistent with a greater effect of tobacco smoke on late-stage carcinogenesis. Similar results were obtained in age- and control site-specific analyses. Further, the age-specific data suggested a dose-response relation between duration of smoking and invasive cervical cancer. An association between alcohol consumption and invasive cervical cancer was not observed.