1. Estimating the relative importance of factors affecting key vital rates is an essential challenge for population ecology. In spite of a large literature on individual growth rates of north temperate-zone fishes, relative effect sizes for the wide range of abiotic and biotic factors affecting fish growth are not well characterized, strongly limiting our ability to predict the effects of environmental change on fish populations. 2. We applied generalized linear mixed models to data from a long-term (nine cohorts over 10 years) individual-based (7685 records from 4203 individuals) study of stream-dwelling juvenile Atlantic salmon (Salmo salar) to estimate the relative importance and interactive effects of stream discharge, water temperature and population density on season-specific growth rates. The model explained a large proportion (r(2) = 0.95) of observed variation in growth. 3. Elasticity analysis was used to estimate the relative importance of model variables on growth in length for Atlantic salmon between age 0+ autumn and the end of age 1+ winter. Effects of population density were substantially weaker than effects of discharge and temperature across all seasons. Opposing among-season temperature effects reduced the overall importance of temperature on growth in contrast to discharge, where effects were generally positive among seasons. Consistent among-season effects and a greater range of observed variation combined to increase the effect of discharge on growth compared to temperature. 4. These results suggest that robust predictions of body growth in north temperate stream fishes will need to include season-specific estimates of variation in both abiotic (temperature and discharge) and biotic (density) factors, but that variation in discharge will dominate growth responses.