Abstract: Place learning abilities represent adaptations that contribute also to foraging efficiency under given spatio-temporal conditions. We investigated if this ability in turn constrains decision making in two sympatric vole species: while the herbivorous common vole (Microtus arvalis) feeds on spatio-temporally predictable food resources (e.g. roots, tubers and shoots of plant tubers), the omnivorous bank vole (Myodes glareolus) additionally subsists on temporally unpredictable food resources (e.g. insects and seeds). Here, we compare the spatial reference memory and working memory of the two species. In an automated operant home cage with eight water places, female voles either had to learn the fixed position of non-depletable places (reference memory task) or learn and avoid previously visited water places depleted in a single visit (win-shift task). In the reference memory task, Microtus females required significantly more choices to find all water places, initially performed slightly worse than Myodes females, and displayed slightly lower asymptotic performance. Both species were highly similar in new learning of the same task. In the more complex win-shift task, asymptotic performance was significantly lower in Microtus (72% correct) than in Myodes (79%). Our results suggest that both vole species resemble each other in their efficiency to exploit habitats with low spatio-temporal complexity but may differ in their efficiency at exploiting habitats with temporally changing spatial food distributions. The results imply that spatial ability adjusted to specific food distributions may impair flexible use of habitats that differ in their food distribution and therefore, decrease a species’ chances of survival in highly dynamic environments.