The pace-of-life syndrome (POLS) hypothesis predicts variation in behaviour and physiology among individuals to be associated with variation in life history. Thus, individuals with “fast” POLS should grow faster, exhibit higher metabolism, be more risk prone, but die earlier than “slow” ones. Empirical support is nevertheless mixed and modelling studies recently suggested POLS to vary along selection gradients. Therefore, including ecological variation when testing POLS is urgently needed to understand whether POLS is a fixed or a variable construct. Here, we tested POLS predictions between and within two fish populations originating from different ecological conditions. We observed opposing life histories between populations, characterized by diverse investments in growth, fecundity, and functional morphology under identical laboratory conditions. Yet, a slower life history was, on average, associated with boldness, high activity, and increased metabolic rates. Within populations, among-individual differences in boldness, activity, standard metabolism, and size-at-age were repeatable over ontogeny, indicating that individuals differed consistently from each other. However, correlation structures among POLS traits were not consistent between populations, with POLS observed in the slow-growing but not in the fast-growing population. Our results suggest that POLS traits can evolve independently from one another and that their coevolution depends upon specific ecological processes.