1. Individuals make behavioural decisions by weighing potential advantages and costs (e.g. increased food intake vs. increased risk of predation). When animals change their activities in response to a perceived threat, their energetic input may decline. Marine ecotourism, including whale and dolphin watching, is growing globally and cetaceans perceive interactions with tour vessels as a form of risk. Observable behavioural changes need to be linked to bioenergetic effects to determine the potential population consequences of this disturbance. 2. We developed a theoretical optimal dive model for bottlenose dolphins under three potential types of perceived risk resulting from human interactions at the surface (decreasing instantaneous risk, increasing instantaneous risk and no risk). We compared the predictions of these theoretical models to observed dive cycles of foraging male and female dolphins in the presence and absence of tour vessels. We used mixture models to classify dive types and mixed effects models to analyse changes in the interbreath interval of surface and bottom dives and the frequency of estimated bottom dives. 3. Males significantly increased bottom time and performed fewer bottom dives when boats were present, matching predictions of our theoretical model for perceived decreasing instantaneous risk. In contrast, females significantly decreased bottom times and increased the frequency of bottom dives, matching predictions from the model for perceived increasing instantaneous risk. Therefore, our empirical results suggest differences in the perception of risk between sexes. 4. Synthesis and applications. By comparing theoretical predictions with observed dive data, our study suggests that boat interactions during foraging can cause decreased net energy gain over a foraging bout for both sexes, with females being more impacted. The population under study is currently listed as critically endangered. Understanding whether these predicted energetic impacts affect an individual's vital rates will provide a link to the population-level consequences of this disturbance. Previous analytical approaches have failed to capture the costs associated with disturbance during foraging, leading to management recommendations that only protect animals from increased energetic expenditure. We suggest that the current management scheme should be revised to include foraging areas in order to secure the energy intake of animals.