Using a panel dataset of commercial fisheries in the Gulf of Mexico, we attempt to separately identify the moral hazard and self‐selection effects of property insurance coverage among commercial fishers. We use captains' propensity to take fishing trips under adverse weather conditions as a proxy for their private information; these data are available to us, but not to insurers. We find that vessels with higher long‐term exposure to risk are significantly less likely to be insured, suggesting potential advantageous selection. However, this relationship dissipates once we control for information likely known to the insurer. Finally, using a Heckit estimator, we find evidence of moral hazard: insured captains take more risks at sea.