The ability to quantify animals’ feeding activity and the resulting changes in their body condition as they move in the environment is fundamental to our understanding of a population’s ecology. We use satellite tracking data from northern elephant seals (Mirounga angustirostris), paired with simultaneous diving information, to develop a Bayesian state-space model that concurrently estimates an individual’s location, feeding activity, and changes in condition. The model identifies important foraging areas and times, the relative amount of feeding occurring therein and thus the different behavioral strategies in which the seals engage. The fitness implications of these strategies can be assessed by looking at the resulting variation in individuals’ condition, which in turn affects the condition and survival of their offspring. Therefore, our results shed light on the processes affecting an individual’s decision-making as it moves and feeds in the environment. In addition, we demonstrate how the model can be used to simulate realistic patterns of disturbance at different stages of the trip, and how the predicted accumulation of lipid reserves varies as a consequence. Particularly, disturbing an animal in periods of high feeding activity or shortly after leaving the colony was predicted to have the potential to lead to starvation. In contrast, an individual could compensate even for very severe disturbance if such disturbance occurred outside the main foraging grounds. Our modelling approach is applicable to marine mammal species that perform drift dives, and can be extended to other species where an individual’s buoyancy can be inferred from its diving behavior.