Native and introduced wild deer have significant unwanted impacts in many countries. Lethal control, usually by hunting, is the most cost-effective method of reducing their number and impact. However, deer habitat use varies spatially and temporally, meaning that hunters (the predator) may search in habitats with few or no deer. Also, deer may modify their behaviour in response to hunting pressure in ways that decrease the risk of being killed, reducing the efficacy of lethal control programmes. To address these issues we decomposed the predation process into its four constituent stages–prey occurrence, predator search, predators encounter prey, predators kill prey they encounter–to reveal what makes female sika deer (Cervus nippon) in the North Island, New Zealand, more susceptible to control by professional helicopter-based hunters. Female sika deer were encountered more in some habitats or landscape features than in others, but the likelihood of kill given an encounter in those habitats was not always correspondingly high. Similarly, variables that positively influenced the probabilities of encounter and kill did not similarly influence deer occurrence, and variables most likely to be searched often poorly correlated with the other predation stages. These disparities contributed to suboptimal searching for and killing of deer by helicopter-based hunters during winter and spring, but not summer. Our study demonstrates how decomposing predation risk can identify risk and refuge areas for target species, and can be used to alter tactics as target species adjust to changing predation risk. This analytical approach is highly applicable to other human hunter–prey systems.