risk allocation theory predicts that foraging animals moderate predation risk by allocating their foraging effort and space use according to their energetic demands. undernourished animals have a greater need for energy than do sated individuals and should accept higher risk while foraging. original theory predicted that the proportion of time that individuals spend in good conditions primarily determines risk allocation. more recent theory predicts that the length of exposure to good or bad conditions governs risk allocation decisions when patterns of environmental risk are autocorrelated in time. i investigate the effects of these factors with controlled experiments on a standard arthropod (folsomia candida). i subjected animals to nine temporally autocorrelated 16- day feeding treatments varying in both the proportion (0, 0.25, 0.50 and 0.75) and duration (short, medium and long intervals) of time when food was absent. risk allocation was assessed by the choice of occupying a risky dry habitat where food was present (rich) versus a safe moist habitat with no food (poor). irrespective of autocorrelation in conditions, the proportion of time spent with no food primarily determined risk allocation by these collembolans. the results suggest an energetic threshold below which f. candida are forced to forage in rich and risky habitat despite the possibility of mortality through desiccation. state dependent patterns of habitat selection suggest that understanding the relationship between energetic state and patterns of environmental condition may allow us to employ risk allocation as a leading indicator of habitat change.