This study evaluated the effect of training on improving the reliability of the peer review process by determining whether or not training made a difference in the variability among reviewers' scores and documentation provided to support the scores. Different levels of training were provided for participants in the peer review process who were reviewing applications for discretionary grants in the Training Personnel for the Education of the Handicapped program. Reviewers' total scores, total word count for strengths and weaknesses, and approval/disapproval decisions were then examined. There were no significant differences in applications' average total scores across the three levels of training, although standard deviations were higher with more training. Also, there was no reduction in the dispersion of the scores with an increase in training. Reviewers who received additional training tended to write more words in both the strengths and weaknesses sections. Concerning approval or disapproval of the application, there were statistically significant differences across the training conditions, but they could not be determined to be training differences. Reviewers across all training conditions provided more documentation for those applications they disapproved than for those they approved. More experienced reviewers tended to give lower scores and more documentation than less experienced reviewers. (JDD)