Purpose: We estimated the effect of a voucher benefit on the demand for personal assistance by Medicare beneficiaries aged 65 years or older who had functional disabilities. Design and Methods: We performed a secondary data analysis on 645 Medicare beneficiaries from the Medicare Primary and Consumer-Directed Care Demonstration (a randomized controlled trial) between August 1998 and June 2000. We estimated a two-part model to determine the effect of the voucher on out-of-pocket personal assistance expenditures. The model controlled for individual health and functional status variables, sociodemographics, prior health care utilization, and state fixed effects. Results: A modest experimental Medicare personal assistance voucher benefit (that reimbursed 80% of up to $250 of eligible expenses per month) increased the likelihood of any out-of-pocket spending for assistance (by 12%, p less than 0.05), but it did not increase the amount of personal assistance expenditures among users (p = 0.94). Overall, the voucher benefits increased average annual expenditures by 10% ($5,304 for the voucher group vs $4,836 for the control group). However, this effect did not reach statistical significance (p = 0.66). Implications: The voucher benefit provided a small incentive to use personal assistance for older Medicare beneficiaries with functional disabilities. Thus, if Medicare were to implement such a benefit, Medicare expenditures may increase. Further research is needed to determine if the increased personal assistance use leads to better health outcomes and whether it is associated with offset cost savings for Medicare-covered services. (Contains 4 tables.)