Purpose: To examine the impact of an experimental consumer-choice voucher benefit on the selection of independent and agency personal assistance services (PAS) providers among rural and urban Medicare beneficiaries with disabilities. Methods: The Medicare Primary and Consumer-Directed Care Demonstration enrolled 1,605 Medicare beneficiaries in 19 counties in New York State, West Virginia, and Ohio. A total of 839 participants were randomly assigned to receive a voucher benefit (up to $250 per month with a 20% copayment) that could be used toward PAS provided by either independent or agency workers. A bivariate probit model was used to estimate the probabilities of choosing either type of PAS provider while controlling for potential confounders. Findings: The voucher was associated with a 32.4% (P less than 0.01) increase in the probability of choosing agency providers and a 12.5% (P = 0.03) increase in the likelihood of choosing independent workers. When the analysis was stratified by rural/urban status, rural voucher recipients had 36.8% higher probability of using independent workers compared to rural controls. Urban voucher recipients had 37.1% higher probability of using agency providers compared to urban controls. Conclusions: This study provided evidence that rural and urban Medicare beneficiaries with disabilities may have very different responses to a consumer-choice PAS voucher program. Offering a consumer-choice voucher option to rural populations holds the potential to significantly improve their access to PAS.