Despite recent advances in the study of dispersal and migration of insects (see reviews by Dingle 1972; Kring 1972; Johnson 1969), knowledge of Drosophila, one of the most important population genetic tools, is based upon released flies recaptured upon artificial baits (Dobzhansky and Wright 1943; Burla et al. 1950; Crumpacker and Williams 1973). Baits are used both to attract individuals for recapture and to impose a uniform recapture pattern on an uneven habitat. Placement of the baits is generally independent of the habitat, and the location and abundance of natural breeding sites are unknown. Highly structured experiments allow elegant analysis of the data based upon the assumptions that (1) interaction between baited attractive areas and naturally attractive breeding sites is relatively unimportant, (2) accumulation of flies at naturally attractive areas produces perturbations to random dispersal, and (3) differences in dispersal rate between fed and unfed flies in attractive and unattractive areas are negligible. The essential role of dispersal, in particular of Drosophila dispersal, in population genetics and population biology makes it important to evaluate the validity of baited release-recapture studies. The baited mark-recapture of the cactiphilic Drosophila nigrospiracula Patt. and Wheeler reported here led to dispersal estimates which were consistent with published results for other Drosophila species. However, the ecology of D. nigrospiracula is well understood. Consideration of the substrate density and natural behavior of these flies shows that the dispersal estimated by the trap technique is too low to permit propagation of the species. Further, the estimates are far below the known dispersal capacity of the species. An analysis of the errors inherent in the study shows that the same errors are present in previous mark-release studies based upon baited recapture. We suggest that the dependence of dispersal rate on habitat and experimental design has been underestimated. We suggest that different experimental designs produce correspondingly different rates of dispersal and that baited studies have tended to obscure, rather than clarify, habitatdispersal relationships.