We compare rates and days of maternity and nonmaternity hospital admission for the years 1981 through 1984 for three groups of employees and dependents from a large private employer: those continuously enrolled in a fee-forservice (FFS) plan (N=147,700), those continuously enrolled in a health maintenance organization (HMO) (N=30,957), and those switching from the FFS plan to the HMO (N=2,144). The rate of maternity admissions for plan switchers increased by 106%(P<0.001) in the post-switch year compared with the pre-switch year, while maternity rates for continuing FFS-plan enrollees declined by 12%(P<0.001) and rates for continuing HMO enrollees remained unchanged. Nonmaternity admission rates for switchers decreased by 19%(P=0.079),consistent with the expectation that HMOs reduce these rates substantially, while rates for FFS-plan stayers increased 4%(P<0.001) and those for HMO stayers remained unchanged. We conclude that employees often switch health plans when anticipating increased needs for maternity care and therefore that pre-switch rates of utilization are unreliable measures of the true magnitude of risk selection between HMOs and FFS plans.