The optimality of behaviour has been the focus of long-standing debate in ecology and evolution. Individuals often appear to make suboptimal choices, e.g. during oviposition or foraging, and this is often attributed to neurological or evolutionary constraints. However, behavioural optimality may also change across environmental contexts in which it is measured. In this study, we measured the context-dependence of oviposition preference and its fitness outcomes for individual Tribolium castaneum females encountering a habitat with two different resources. We first estimated the fitness outcomes of all possible oviposition choice strategies, by manipulating egg distribution between patches of the optimal resource and a novel suboptimal resource. To test the impact of different environmental contexts experienced by a reproductive female on her oviposition choice, we measured oviposition preference of young vs. old females that had experienced either low or high population density. We found that depending on the experienced context, females optimize different aspects of offspring fitness. Females exposed to high density maximized offspring development rate. In contrast, young 'naive' females oviposited indiscriminately, which appeared to be the worst strategy because it greatly delayed offspring development. However, when we extended the duration of the experiment, these females ultimately produced as many or more adult offspring than other female groups, suggesting that this strategy may not be maladaptive when colonizing new areas. Thus, apparently maladaptive oviposition choices may actually reflect incomplete information on context-dependent fitness outcomes. Our results underscore the importance of evaluating behavioural optimality across different environmental contexts, and suggest that ignoring context-dependence may lead to biases in inferences about the ecological and evolutionary processes that drive behavioural choice.