Phenotypic plasticity is one major source of variation in natural populations. Inducible defences, which can be considered threshold traits, are a form of plasticity that generates ecological and evolutionary consequences. A simple cost–benefit model underpins the maintenance and evolution of these threshold, inducible traits. In this model, a rank-order switch in expected fitness, defined by costs and benefits of induction between defended and undefended morphs, predicts the risk level at which individuals should induce defences. Here, taking predator-induced morphological defences in Daphnia pulex as a threshold trait, we provide the first comprehensive investigation into the costs and benefits of a threshold trait, and how they combine to reflect fitness and predict the switchpoint at which induction should occur. We develop reaction norms that show genetic variation in switchpoints. Further experiments show that induction can confer a survival benefit and a cost in terms of lifetime reproductive success. Together, these two traits combine to estimate expected fitness and can predict the switchpoint between an undefended and a defended strategy. The predictions match the reaction norm data for clones that experience these costs and benefits, and correspond well to independent field data on induction. However, predictions do not, and cannot, match for clones that do not gain a benefit from induction. This study confirms that a simple theory, based on life history costs and benefits, is a sufficient framework for understanding the ecology and evolution of inducible, threshold traits.