It is often stated that people are irrational or more specifically that they fail to correspond to one or other rational norm. How- ever, a broader consideration of the constraints to which people adapt sometimes reveals rationality and thereby provides ex- planations of human behavior. For example, Hahn and Warren (2009) showed that a consideration of the nature of people’s actual experience of a fair coin toss can help explain seem- ingly biased perceptions of randomness. In this paper we re- port a study aiming to reveal the extent to which human choices about sampling are rational. The paper contrasts a model of sampling based on individual experience with one based on the payoff statistics of the local task environment. The result indicates that the number of samples that participants choose is better explained by a rational response to individual experi- ence than by what appears rational given the statistics of the local task environment.