Thirty-eight college students were the subjects of a study designed to examine the efficacy of training graduate reading specialists through a diagnosis simulation program versus the use of a traditional child study approach. The major task of this investigation was to compare, by way of univariate or multivariate analysis, various combinations of simulation program-child study approaches. The sequential, hierarchical, and systematically structured computer simulation of human cognition (Simon, 1979) formed the basis of the present simulation program. It was expected that those students involved in some aspect of the simulation would demonstrate superior skills by scoring higher on a post-treatment measure of diagnostic abilities than students involved only in traditional child studies. It was found that students in the child study simulation group scored significantly higher than students in the child study group on only parts one and four of the final exam. This finding supported the contention that a specific form of simulation can benefit students who are being trained to do reading diagnosis, and can serve as a useful adjunct to the traditional practicum experience. (MER)