Using data on women’s choice of hospital for childbirth in Florida, we find that women return to the same hospital approximately 70% of the time. We separate explanations of switching costs and unobserved preference heterogeneity using a panel data fixed effects estimator and find that switching costs account for approximately 40% of the demand effects of a lagged dependent variable. The welfare effects of excluding a hospital from a payer’s network are smaller in the short run but higher in the long run, given our estimates of switching costs, and the dynamic effects of entry on competition are significantly smaller.