Researchers taking a domain-general approach to the development of scientific reasoning long thought that the ability to engage in scientific reasoning did not develop until adolescence. However, more recent studies have shown that preschool children already have a basic ability to evaluate evidence and a basic understanding of experimentation. Data providing insights into when exactly in the preschool years significant gains in these abilities occur are scarce. Drawing on a sample of 138 preschool children, this longitudinal study therefore examined how children's ability to evaluate evidence and their understanding of experimentation develop between the ages of four and six. Findings showed that the ability to evaluate evidence was already well developed at age four and increased steadily and significantly over time as long as the pattern of covariation was perfect. In the case of imperfect covariation, the proportion of correct answers was low over the period of observation, but showed a significant increase between the ages of four and five. If the data did not allow relationship between variables to be inferred, the proportion of correct answers was low, with a significant increase between the ages of five and six. The children's understanding of experimentation increased significantly between the ages of five and six. The implications of these findings for age-appropriate science programs in preschool are discussed.