The food habits fo Microtus pennsylvanicus in southeastern Manitoba were studied in conjunction with an analysis of the plant community existing on the study plot. Snap-trapping, laboratory food preference tests, and examination of the stomach contents were employed to determine local distribution of the animals and their food preferences. Both the preference tests and stomach analysis showed that certain plant species were highly preferred. Species which were preferred both in the laboratory and in the wild included Bromus inermis, Taraxacum officinale, species of Carex and Melilotus, and Trifolium repens. Underground stems, roots, underground fungi, and mosses were also eaten frequently. Multiple regression analysis indicated an association between Microtus and species of Poa, which were not, however, a preferred food. The degree to which Poa was associated with some preferred plant species suggested that the association of voles with Poa may reflect the ability of these preferred foods to coexist with Poa. The animals may choose this habitat because it affords both cover and preferred foods. Vole numbers were not correlated with good cover as provided by a species (such as Calamagrostis inexpansa) which was not readily eaten and did not occur in association with more palatable plants. The evidence of selective feeding and the indication that voles tend to be associated with certain plant associations suggested that food preferences, and, to some extent, cover, affect local distribution and perhaps migration and population levels.