Qualitative and microcosm models were used to predict effects of herbicide application for control of Myriophyllum spicatum. Predictions were compared to data from Pat Mayse Lake, a Texas reservoir, where localized areas were treated with endothall. Although milf oil was temporarily eliminated, when endothall was used according to manufacturer's directions, no ecologically significant direct or indirect effects were observed on nontarget species or abiotic water quality. Comparisons of the predictions with field data confirmed the capabilities of this approach for estimating risk and emphasizing the importance of identifying regulating or driving factors that modify environmental impacts of aquatic weed control programs so they can be incorporated into future risk assessments.