The Behavioral Risk Factor Surveillance System (BRFSS) is widely used by state health agencies to measure the prevalence of chronic disease risk factors. We completed a test-retest study to assess the reliability of the Missouri Behavioral Risk Factor Surveillance System. We conducted telephone reinterviews for 222 respondents of completed Behavioral Risk Factor Surveillance System interviews from March and April 1993. The second interview was completed between 6 and 30 days after the first interview. Agreement was high for sociodemographic variables (kappa values from 0.85 to 1.00). Reliability of information on chronic conditions and risk factors was also high, with kappa values from 0.82 for hypertension to 1.00 for current smoking status. Regarding cancer screening practices, reliability was lower for knowledge of the prostate-specific antigen test (kappa = 0.21) than for women's cancer screening practices (that is, the mammogram and Papanicolaou smear). Questions on attitudes toward environmental tobacco smoke showed lower reliability than did questions on individual actions to reduce exposure to environmental tobacco smoke.