1. While fish-based Indices of Biotic Integrity (IBIs) have been developed for a wide array of lotic systems, equivalent tools have seldom been adapted to the monitoring and assessment of lakes. Major difficulties arise in such work: collecting data that allow statistically robust analyses, choosing the relevant variables to describe the biotic, environmental and anthropogenic data sets and assessing the relative importance of the latter two in explaining the former. The aim of the present paper is to produce such an assessment for the fish communities of the lakes of north-east USA. 2. Fish surveys, environmental features and catchment-scale descriptors of human stresses(agricultural and urban land-uses) were collected for 112 natural lakes. 3. Fish metrics, i.e. species richness and percentages of species belonging to reproductive, trophic, and tolerance guilds, were regressed against anthropogenic variables, then against anthropogenic variables and the natural environmental conditions.4. It was shown that failing to control for the natural environmental conditions in the IBI construction led to selecting metrics (percentage of intolerant species and percentage of omnivorous species) that did not display response to stresses when the environment was controlled for. Moreover, controlling for natural variability of the metrics allowed identifying the impact of agricultural land-use on the percentage of diadromous species.5. Fish communities appear valuable for the bioassessment of lakes. Appropriate statistical methods have proved that the natural variability in the bioassessment tools could be accounted for, thereby allowing assessments at the scale of multiple basins and ecoregions. This opens new perspectives for the development of IBIs for lentic systems in lake-poor regions, such as southern Europe, and therefore represents a significant contribution to the implementation of the European Water Framework Directive.