Ecologists utilize many approaches to assess risk of invasive species establishment. We subscribe to an approach that first considers introduction effort to identify which lakes are exposed to an invader, following which the environmental suitability and ecological integration ability are considered. We highlight this approach using all three steps for the zebra mussel in Ontario. We measured introduction effort using a 'gravity model', which linked invaded source with noninvaded destination lakes via an array of human activities. Of 291 lakes, invasion status was accurately predicted for 80% using only gravity scores in a Multivariate Adaptive Regression Spline analysis; 69% of 48 lakes predicted to be invaded were invaded. Another MARS model based upon physical and chemical parameters indicated that establishment was correlated with two parameters: dissolved organic carbon (-) and pH (+). Of lakes predicted to be invaded based upon introduction effort, 75% were correctly classified with regard to invasion status when lake characteristics were considered. Of these lakes, 94% predicted to be invaded were invaded. Another MARS model used fish community composition for lakes predicted to be invaded from the previous models; 30 of 35 lakes predicted to be invaded were invaded. Establishment was negatively correlated with presence of whitefish, which prey on Dreissena, and positively correlated with largemouth bass, reflecting predation release from molluscivorous fishes by the bass. While this approach is data and computationally intensive, it may provide the most realistic assessment of where spreading nonindigenous species may colonize successfully.