Stock price behavior is often described via the market model, in which the residual term is presumed to measure abnormal returns available to the investor. This error term is assumed to: (1) have a mean of zero and (2) be normally distributed, with a constant variance. Empirical tests, however, sometimes report abnormal returns, whose distributions violate the assumptions which underlie them. The research reported here argues that such deviations might be explained by the presence of a bimodal error term. The existence of bimodality would help explain those observations in which: (1) cumulative abnormal returns are not zero and (2) levels of confidence are lower than expected. In addition, the bimodal model helps explain the variability of post-event adjustment lag.