AbstractAssortative mating is thought to play a key role in reproductive isolation. However, most experimental studies of assortative mating do not take place in multiple natural environments, and hence, they ignore its potential context dependence. We implemented an experiment in which two populations of brown trout (Salmo trutta) with different natural flow regimes were placed into semi‐natural stream channels under two different artificial flow regimes. Natural reproduction was allowed, and reproductive isolation was measured by means of parentage assignment to compare within‐population vs. between‐population male–female mating and relative offspring production. For both metrics, reproductive isolation was highly context dependent: no isolation was evident under one flow regime, but strong isolation was evident under the other flow regime. These patterns were fully driven by variance in the mating success of males from one of the two populations. Our results highlight how reproductive isolation through assortative mating can be strongly context dependent, which could have dramatic consequences for patterns of gene flow and speciation under environmental change.