Living under predation risk may alter both behaviour and physiology of potential prey. In extreme cases, such alterations may have serious demographic consequences, and recent studies support that non-lethal effects of predation may have broad ecological consequences. However, behavioural and physiological responses to predation risk may be related to trade-offs associated with resource acquisition and direct predation risk. We validated an enzyme-linked immunoassay (EIA) for non-invasive monitoring of stress in plains zebras (Equus quagga) from faecal material. We used this assay in combination with behavioural data to assess if plains zebras living with and without lions (Panthera leo) in a mountain savannah in southern Africa differed in behaviour and physiology, and if such differences were influenced by seasons with contrasting resource availability. Zebra group sizes did not differ between areas with and without lions, but zebra groups had more juveniles in an area with lions than groups in an area without lions, but only during the wet season. Similarly, we observed differences in individual vigilance, foraging behaviour and stress hormone concentrations, but all these differences were influenced by seasons. Despite these seasonal influences, our study did not suggest that zebras in an area with lions spent a higher proportion of time being vigilant, a lower proportion of time foraging, or had higher stress hormone levels. Our results instead suggest that zebras' responses to lion presence were highly context dependent and the result of complex interactions between resource abundance and cues about predation risk. Because of the obvious ecological and evolutionary ramifications of such findings, we argue that further research is needed to define the spatial and temporal scales over which predators impose indirect effects on their prey.