In the absence of adequate risk adjustment, capitation for enrollees creates incentives for health plans to enroll and retain good risks and to avoid bad risks. This article examines whether Maryland Medicaid beneficiaries with histories of substance abuse disenroll from health plans more frequently than those without such histories. The findings indicate that enrollees with a history of substance abuse were more likely to switch plans than other enrollees, regardless of whether they chose the health plan or were randomly assigned to the plan. These results suggest that current risk-adjustment systems may fail to offset selection incentives in modern capitated health plans.