A complete understanding of life history evolution requires an appreciation for the influence of natural selection and genetic variation on intrapopulational phenotypic variation in key life history traits. Experimental manipulation provides particularly useful insight into the biological importance of this variation. Consequently, I undertook a field experiment in which I manipulated body size and evaluated selective predation in order to investigate the ecological and evolutionary significance of phenotypic variation in hatch- ling turtles. I incubated eggs from 17 clutches of common snapping turtles (Chelydra serpentine) on wet and dry substrates, producing "large" and "small" hatchlings, respec- tively. Hatchlings were individually marked, measured, and then released simultaneously at a typical nest site in a National Wildlife Refuge area, Whiteside County, Illinois from which the eggs were collected originally. Subsequently, hatchlings were recaptured in drift fences erected along the edge of the Mississippi River. Sixty-six of 112 hatchling C. serpentine were recaptured. Survivorship was not related to clutch, incubation conditions, or locomotor performance, but was significantly size dependent. Natural selection favored larger hatchlings. The strength of selection acting directly on three size traits, as measured by standardized selection gradients (d'), was 0.327 (PCI [principal component 1] = overall size and mass), 0.282 (PC3 = plastron length), and 0.162 (PC4 = trade-off between hatchling mass and width). The opportunity for selection was moderate (I = 0.703), but did not constrain selection. Full-sib heritabilities of body size of hatchling turtles were variable (h2We, = 0.72 ? 0.34 and h2dr = 0.17 ? 0.28), but were not significantly different between the two moisture treatments. These results provide support for the "bigger is better" hypothesis. Larger hatchling Chelydra serpentine exhibited significantly greater survivorship than smaller individuals during movement from the nest site to water. However, larger body size of hatchling turtles may not evolve rapidly because the strength of selection was moderate in magnitude and the heritability was relatively low. These results emphasize the need for more detailed ecological investigations of the early life history of these organisms. Furthermore, this study highlights the utility of experimental manipulations of phenotypic variation for evaluating the potential impact of natural selection on life history evolution.