Health care utilization and costs under managed care are evaluated in this report. As part of the Civilian Health and Medical Program of the Uniformed Services (CHAMPUS) Reform Initiative (CRI) demonstration, conducted from 1988 to 1993 in California and Hawaii, this report includes estimates of the effects of CRI on utilization levels and costs for CHAMPUS beneficiaries. It also explores differences within CRI between beneficiaries who elected to enroll in CHAMPUS Prime, an HMO option offered by CRI, and those who did not enroll. The authors conducted two mail surveys of CHAMPUS beneficiaries, (beneficiaries include dependents of active-duty military personnel, and retired military personnel and their dependents) one shortly before CRI began and another two years later. They then compared data from both surveys in the eleven military hospital catchment areas with eleven matched control areas in different states. The authors found that for the average adult beneficiary, costs to the government were 9 percent higher with CRI. Compared to the non-CRI program, costs were 57 percent higher for Prime enrollees, whereas they were the same for non-enrollees. Prime enrollees' use of outpatient care accounted for almost all of the utilization increase in CRI. Active-duty spouses who enrolled did not change their military treatment facility (MTF) use, but they were more likely to augment their (MTF) care with civilian care. Retired enrollees were more likely to use both MTF care and civilian care. CHAMPUS inpatient utilization was lower in CRI, as is often the case in managed-care programs, whereas use of MTF inpatient services did not change significantly. These findings suggest that CRI was able to increase access, especially to civilian care, with an accompanying increase in costs. The evidence points to high utilization among Prime enrollees, especially for retired beneficiaries. First-dollar coverage in Prime increased the costs of care that would have been used even without CRI, thereby adding to the amount of care demanded. The cost containment features in CRI, such as utilization review, were not able to counteract the added costs in Prime and the higher administrative overhead for the program.