We develop a test for adverse selection and use it to examine private health insurance markets. In contrast to earlier papers that consider a purely private system or a system in which private insurance supplements a public system, we focus our attention on a system where privately funded health care is substitutive of the publicly funded one. Using a model of competition among insurers, we generate predictions about the correlation between risk and the probability of taking private insurance under both symmetric information and adverse selection. These predictions constitute the basis for our adverse selection test. The theoretical model is also useful to conclude that the setting that we focus on is especially attractive to test for adverse selection. Using the British Household Panel Survey, we find evidence that adverse selection is present in this market.