With the growing interest in research on expertise, a diverse set of research approaches has emerged. Experts are often identified using social criteria and the amount of experience and the level of performance of these experts is then compared to those of novices and less experienced subjects on various tasks. In this paper, we argue that the success of the pioneering research on chess was due to the satisfaction of a number of additional constraints. The primary focus of the pioneering research on expertise in chess was to study stable superior performance in real life expert activities. Hence the first step in such a study involves an analysis of that real life performance to identify a set of tasks which captures reliably the superior performance of experts in the laboratory. Once the superior performance of experts has been reliably reproduced in the laboratory, examination of the critical mediating processes responsible for the superior performance can be made using standard methods of process tracing and experimental analysis. When the critical processes have been identified, the issue of how these processes can be acquired during extended practice is raised. Theoretical analyses of learning and skill acquisition are explored. Most importantly, empirical studies of the acquisition of the critical processes can be made by cross-sectional comparisons of subjects with differing amounts of expertise and experience, by studies of extended training on the target tasks and by a careful analysis of the real-life practice activities designed to maximize improvement of performance.