We find that asymmetric information is important for the uptake of supplementary private health insurance and health care utilization. We use dynamic panel data models to investigate the sources of asymmetric information and distinguish short-run selection effects into insurance from long-run selection effects. Short-run selection effects (i.e. responses to shocks) are adverse, but small in size. Also long-run effects driven by differences in, for example, preferences and risk aversion, are small. But we find some evidence for multidimensional asymmetric information. For example, mental health causes advantageous selection. Estimates of health care utilization models suggest that moral hazard is not important. © 2012 Elsevier B.V.