The selection response of the polymorphic host D. melanogaster (Meigen) to the braconid wasp A. tabida (Nees) is addressed. Cages of flies with and without wasps were initiated with a population of D. melanogaster that exhibited variation both in larval foraging behavior and in encapsulation ability. Encapsulation ability was measured as the proportion of parasitized larvae that produce a hardened capsule which encapsulates the wasp egg and ultimately kills the wasp larva. We determined whether the host population changed its encapsulation ability and/or its foraging behavior in response to the wasp. Both species were collected from a local orchard where A. tabida is the only wasp known to parasitize D. melanogaster larvae. The naturally occurring genetic polymorphism for rover and sitter larval foraging behavior in D. melanogaster is also found in this field population. A. tabida's vibrotactic search behavior enables it to detect rover more frequently than sitter larvae. Rover larvae move significantly more while feeding than do sitter larvae. In this field population, rover larvae also show higher encapsulation abilities than do sitter larvae. Six cage populations, three without wasps and three with wasps, each containing an equal mixture of rover and sitter flies, were established in the laboratory and maintained for 19 fly generations. Selection pressure in the laboratory was similar to that found in the field population from which the flies and wasps were derived. We found that larvae from cages with wasps developed a significantly higher frequency of encapsulation than those reared without wasps. We were, however, unable to detect a change in larval movement (rover or sitter behavior) in larvae from cages subject to selection from wasps compared to larvae from cages containing no wasps. This may have resulted from a balance between two selective forces, selection against rovers by the wasps' use of vibrotaxis, and selection for rovers resulting from their increased encapsulation abilities.