We used depletable food patches to determine the effect of microhabitat (mowed versus unmowed adjacent grasslands) and time (day versus night) on the foraging behavior of common voles (Microtus arvalis). The food remaining after 12-h periods (giving-up density, GUD) measured the vole's habitat selection under predation risk. In accord with several other rodent species and the effects of avian predators, voles had significantly lower GUDs in the unmowed than mowed portion of the grassland. GUDs in patches along the border between adjacent habitats were more similar to the risky mowed grassland than the safe unmowed grass. Time interacted strongly with microhabitat. In the mowed grass, voles had significantly higher GUDs at night than day. Whereas in the unmowed grass, GUDs were significantly higher during the day than night. Vole GUDs did not vary with time along the boundary. This suggests that predators are more abundant or effective in the mowed grass at night (owls?), and in the unmowed grass during the day (weasels?). In terms of predation risk, the voles perceived the mowed grass at night as the riskiest and the unmowed grass at night as the safest. Voles may have difficulties assessing resources under high predation risk: GUDs among patches were well equalized in the unmowed microhabitat whereas in the mowed grass only day GUDs did not vary significantly among patches. We linked these results to the vole's day-night-activity and life span. For the 533 voles live-trapped at the study area, the ratio of day versus night captures for each individual served as an activity index and the span between first and last capture measured minimum life span. In accord with higher GUDs at night, very few individuals behaved selectively towards the night, but individual life expectancy increased with temporal opportunism. Microhabitat differences in GUDs reflect short-term strategies of predator avoidance and the trapping data reflect long-term patterns of anti-predator behavior.