As a key life-history trait, growth rates are often used to measure individual performance and to inform parameters in demographic models. Furthermore, intraspecific trait variation generates diversity in nature. Therefore, partitioning out and understanding drivers of spatiotemporal variation in growth rate is of fundamental interest in ecology and evolution. However, this has rarely been attempted owing to the amount of individual-level data required through both time and space, and issues with missing data in important covariates. Here, we implemented a Bayesian state-space model using individual-level data from 20 populations of Arctic charr (Salvelinus alpinus) across 15 capture occasions, which allowed us to: (i) integrate over the uncertainty of missing recapture records; (ii) robustly estimate size-dependence; and (iii) include a covariate (water temperature) that contained missing data. Interestingly, although there was substantial spatial, temporal and spatiotemporal variation in growth rate, this was only weakly associated with variation in water temperature and almost entirely independent of size, suggesting that spatiotemporal variation in other environmental conditions affected individuals across sizes similarly. This fine-scale spatiotemporal variation emphasizes the importance of local conditions and highlights the potential for spatiotemporal variation in a size-dependent life-history trait, even when environmental conditions are apparently very similar.