The existence of adaptive phenotypic plasticity demands that we study the evolution of reaction norms, rather than just the evolution of fixed traits. This approach requires the examination of functional relationships among traits not only in a single environment but across environments and between traits and plasticity itself. In this study, I examined the interplay of plasticity and local adaptation of offspring size in the Trinidadian guppy, Poecilia reticulata. Guppies respond to food restriction by growing and reproducing less but also by producing larger offspring. This plastic difference in offspring size is of the same order of magnitude as evolved genetic differences among populations. Larger offspring sizes are thought to have evolved as an adaptation to the competitive environment faced by newborn guppies in some environments. If plastic responses to maternal food limitation can achieve the same fitness benefit, then why has guppy offspring size evolved at all? To explore this question, I examined the plastic response to food level of females from two natural populations that experience different selective environments. My goals were to examine whether the plastic responses to food level varied between populations, test the consequences of maternal manipulation of offspring size for offspring fitness, and assess whether costs of plasticity exist that could account for the evolution of mean offspring size across populations. In each population, full-sib sisters were exposed to either a low- or high-food treatment. Females from both populations produced larger, leaner offspring in response to food limitation. However, the population that was thought to have a history of selection for larger offspring was less plastic in its investment per offspring in response to maternal mass, maternal food level, and fecundity than the population under selection for small offspring size. To test the consequences of maternal manipulation of offspring size for offspring fitness, I raised the offspring of low- and high-food mothers in either low- or high-food environments. No maternal effects were detected at high food levels, supporting the prediction that mothers should increase fecundity rather than offspring size in noncompetitive environments. For offspring raised under low food levels, maternal effects on juvenile size and male size at maturity varied significantly between populations, reflecting their initial differences in maternal manipulation of offspring size; nevertheless, in both populations, increased investment per offspring increased offspring fitness. Several correlates of plasticity in investment per offspring that could affect the evolution of offspring size in guppies were identified. Under low-food conditions, mothers from more plastic families invested more in future reproduction and less in their own soma. Similarly, offspring from more plastic families were smaller as juveniles and female offspring reproduced earlier. These correlations suggest that a fixed, high level of investment per offspring might be favored over a plastic response in a chronically low-resource environment or in an environment that selects for lower reproductive effort.