In the context of conservation hatcheries that seek to bolster wild populations by releasing captively‐reared fishes into the wild, steelhead Oncorhynchus mykiss were used to test the hypothesis that naturalistic rearing environments promote adaptive behaviour that might otherwise not develop in typical hatchery environments. When comparisons were made among fish reared in barren, structured or structurally variable environments (i.e. the location of the structure was repositioned every 2–3 days), structure in the rearing environment increased future exploratory behaviour, but only if the structure was stable. Under conditions of high perceived predation risk, the fish no longer exhibited increased exploratory behaviour, suggesting that it is expressed in an adaptive, context‐dependant manner. Another concern with hatcheries is that relaxed selection over multiple generations in captivity can increase maladaptive behavioural variation. Compared to rearing in hatchery‐typical barren environments, rearing in structured‐stable environments decreased behavioural variation. This effect, which occurred during development and did not involve selection, demonstrates a different mechanism for change in behavioural variation in captivity. These experiments show that effects of structure and structural stability occur at the level of both average behaviour and behavioural variation, and suggest that these effects should be considered when fishes are reared in hatcheries for later release into the wild.