Oviposition preference for urea-supplemented food was assayed by simultaneous choice trials on five pairs of closely related laboratory populations of Drosophila melanogaster. Each pair of populations had been derived from a separate ancestral population about 85 generations prior to this study. One population in each pair had been subjected to selection for larval tolerance to the toxic effects of urea; the other population served as a control. Considerable variation in oviposition preference was seen both within and among populations, with four of the ten populations showing a significant mean preference for urea-supplemented food. The degree of specificity shown by individual females was surprisingly high, leading to a bi-modal distribution of oviposition preference in some populations. Overall, selection for larval tolerance to urea did not significantly affect oviposition preference. However, the data indicated that pair-wise comparisons between randomly selected populations from the two larval selection regimes would lead to a range of possible outcomes, suggesting, in several cases, that selection for larval urea tolerance had led to significant differentiation of adult oviposition preference for urea in one or the other direction. The results, therefore, highlight the importance of population level replication and caution against the practice, common in ecological studies, of assaying oviposition preference in two populations that utilize different hosts in nature, and then drawing broad evolutionary inferences from the results.