The purposes of this study were (1) to assess the utility of the economic theory of demand for insurance for modeling voluntary Medicare drug benefit enrollment decisions and (2) to explore the degree of adverse selection and crowd-out that might occur under a voluntary enrollment Medicare prescription benefit. Data were collected using a cross-sectional, mail survey of 2,100 community-dwelling adults aged 65 and older in Wisconsin. Respondents were asked to evaluate their likelihood of enrollment in any of 4 hypothetical drug benefit plans under the assumption that they could enroll in one of the hypothetical plans or maintain their current coverage. Data analyses included bivariate comparisons across enrollment likelihood categories and logit analysis of enrollment likelihood as a function of respondent characteristics. 1041 usable survey forms were returned for an adjusted response rate of 51.5%. Older adults with 4 or more chronic conditions were most likely to report that they were "very likely" to enroll in one of the hypothetical drug plans, as were those with the highest out-of-pocket drug spending in the previous 30 days. Respondents with no or self-purchased drug benefits were more likely than those with employer-based plans to express a higher likelihood of enrollment in one of the hypothetical plans. Adverse selection may be problematic for a voluntary enrollment Medicare drug benefit. Given that high out-of-pocket drug spending (secondary to drug coverage source) was a consistent predictor of enrollment likelihood, demand-side factors affecting the crowding out of employer-based drug coverage sources by a voluntary enrollment drug benefit appear minimal. However, the availability of a Medicare prescription benefit may still lead to crowd-out through employer incentives.