In the context of warming freshwater habitats, protection of Atlantic salmon populations requires an understanding of the effects of temperature on somatic growth during the juvenile life stage. However, quantifying the effect of temperature on growth is challenging given differences among methodologies, metrics of growth, and their underlying assumptions. Using short term studies (2000–2002) in two Canadian populations of wild Atlantic salmon (Margaree and Miramichi rivers), we investigate whether different hierarchical modeling approaches influence the derivation of temperature-growth relationships, by contrasting seasonal growth trajectories (von Bertalanffy; VBGF) to size-at-age data models built with instantaneous growth rates. Size-at-age data analysed seasonally with the VBGF framework failed to detect an effect of temperature, whereas instantaneous growth rates from the same dataset were strongly related to temperature metrics. However, instantaneous growth rates cannot be used to extrapolate predictions into meaningful metrics for fisheries management (e.g., size at the end of the growing season). Nevertheless, we show that size at the end of the growing season can be predicted with VBGF models accounting for site-level variation, which in turn are related to temperature metrics, as observed for instantaneous growth rates. Taken together, these results show that combining these two approaches (size-at-age, growth rates) can circumvent their intrinsic drawbacks and reveal essential ecological patterns that may otherwise remain undetected. In cases where instantaneous growth rates are not available, relating predicted size-at-age from hierarchical VBGF to temperature provides an interesting alternative for detecting subtle environmental effects, even if the VBGF parameters or its residuals are unrelated to temperature metrics.