To examine the suitability of fish scales as potential tracers of nutrient pollution, we analysed the nitrogen and carbon stable isotope values (δ15N and δ13C) in scales of a generalist fish species, roach Rutilus rutilus, collected from 22 Czech reservoirs covering wide gradients of catchment land use and nutrient enrichment. Using generalised additive mixed models in the first step and generalised linear mixed models in the second step, we evaluated the response of roach scale stable isotope values to catchment land use variables (percentage of agricultural land and human population density) and in-reservoir water quality variables. Roach scale δ15N values varied by 15 ‰ among the reservoirs and were strongly, linearly, and positively associated with the percentage of agricultural land in the reservoir catchments, pointing to agriculture as the dominant source of nitrogen pollution in the investigated systems. Roach scale δ13C values differed by 8 ‰ among the studied reservoirs and were not related to catchment land use variables or in-reservoir primary production (chlorophyll-a levels). Possible variation in roach foraging strategies (littoral versus pelagic) between reservoirs or the contrasting effects of eutrophication-related autotrophic and heterotrophic processes on baseline δ13C values may explain the lack of relationships between roach scale δ13C values and the explanatory variables. In summary, our findings show that fish scale δ15N values are sensitive bioindicators of catchment-derived anthropogenic nitrogen inputs to freshwater ecosystems. Because scales can be sampled in a nonlethal way and δ15N analysis is relatively inexpensive, we suggest that measuring the δ15N values of fish scales could be an effective method for monitoring nitrogen pollution in aquatic environments.