We introduce a spatial autoregressive model for cross-sectional correlation of abnormal returns. In the model the abnormal returns of firms in the same industry are correlated, whereas the abnormal returns of firms in different industries are uncorrelated. Tests for abnormal returns which are robust to event-induced cross-sectional correlation are proposed. We apply our tests to U.S. stock returns from Bear Stearns’ collapse and Lehman Brothers’ bankruptcy in 2008. We document evidence of event-induced cross-sectional correlation. Simulations show that tests which estimate the cross-sectional correlation from the event period have size close to the nominal level.