In situ bioassays were performed as part of the Episodic Response Project, to evaluate the effects of episodic stream acidification on mortality of brook trout (Salvelinus fontinalis) and forage fish species. We report the results of 122 bioassays in 13 streams of the three study regions: the Adirondack mountains of New York, the Catskill mountains of New York, and the Northern Appalachian Plateau of Pennsylvania. Bioassays during acidic episodes had significantly higher mortality than did bioassays conducted under nonacidic conditions, but there was little difference in mortality rates in bioassays experiencing acidic episodes and those experiencing acidic conditions throughout the test period. Multiple logistic regression models were used to relate bioassay mortality rates to summary statistics of time-varying stream chemistry (inorganic monomeric aluminum, calcium, pH, and dissolved organic carbon) estimated for the 20-d bioassay periods. The large suite of candidate regressors also included biological, regional, and seasonal factors, as well as several statistics summarizing various features of aluminum exposure duration and magnitude. Regressor variable selection and model assessment were complicated by multicol-linearity and overdispersion. For the target fish species, brook trout, bioassay mortality was most closely related to time-weighted median inorganic aluminum. Median Ca and minimum pH offered additional explanatory power, as did stream-specific aluminum responses. Due to high multicollinearity, the relative importance of different aluminum exposure duration and magnitude variables was difficult to assess, but these variables taken together added no significant explanatory power to models already containing median aluminum. Between 59 and 79% of the variation in brook trout mortality was explained by models employing between one and five regressors. Simpler models were developed for smaller sets of bioassays that tested slimy and mottled sculpin (Cottus cognatus and C. bairdi) as well as blacknose dace (Rhinichthys atratulus). For these forage species a single inorganic aluminum exposure variable successfully accounted for 86-98% of the observed mortality. Even though field bioassays showed evidence of multiple toxicity factors, model results suggest that adequate mortality predictions can be obtained from a single index of inorganic Al concentrations during exposure periods.