Habitat selection by animals that migrate or disperse ultimately determines the biotic and abiotic environment they will experience in subsequent life stages. Intuitively, for habitat selection to be adaptive, animals should respond positively to cues produced by habitat characteristics that will enhance their fitness in the new environment. However, there are many examples of dispersing animals where individuals are attracted to cues produced by factors that reduce their fitness after arrival. In this study, we use a temperate reef fish to examine the relative importance of habitat-associated cues in habitat selection decisions, and assess whether use of these cues is adaptive across early life stages. We used a series of laboratory- and field-based manipulative experiments to test: (1) what habitat-associated cues are likely used to locate suitable habitat; (2) whether in situ settlement patterns reflect the cue response tested in the laboratory; and (3) whether the aspects of the habitat that stimulate settlement are the same as those that maximize survival. We observed a positive response to multiple habitat-associated cues, with conspecific cues eliciting the strongest behavioral response in laboratory choice experiments, and a strong inverse density-dependent relationship at settlement. Macroalgal cues also elicited a positive response at settlement, but were associated with higher mortality after settlement, suggesting that habitat selection decisions are not always adaptive. We argue that this non-intuitive behavior may still be adaptive if it improves fitness at an earlier life stage, as habitat selection behavior is the result of tradeoffs in fitness costs across multiple stages.