Typically, a selection response is lost when a selection regime is relaxed or terminated. Some possible explanations for this result are a negative genetic correlation between selected characters and fitness, antagonistic pleiotropy and linkage involving alleles of opposed effect, or a side-effect of relaxed inbreeding. The current study relaxes selection on stocks selected for resistance to starvation and desiccation stress, and the response to relaxed selection is then observed. Contrary to many findings, we found maintenance of the selection response with respect to some characters upon relaxation of the stress selection regime. Specifically, after 35 generations of relaxed selection, longevity, early fecundity, and desiccation resistance have not changed significantly in the relaxed desiccation-selection populations, suggesting that the alleles affecting these characters lack significant antagonistic pleiotropy. Starvation resistance, on the other hand, rose significantly in the relaxed desiccation populations. After 20 generations of relaxed selection, starvation resistance fell dramatically in the relaxed starvation-selection populations relative to their ancestral populations. Longevity, however, has not dropped significantly from that of its ancestral population. When early fecundity in the relaxed starvation-selection populations is analyzed separately at each generation, there is a significant increase in this character. This increase in early-life fecundity in association with the decrease in starvation resistance may reflect antagonistic pleiotropy between these two characters in the relaxed starvation resistance system.