An item analysis of Harris' scoring system for the Goodenough-Harris Draw-A-Man Test was conducted by comparing sets of protocols obtained in the early fall of kindergarten from children whose overall in-class academic performance placed them either in an at-risk category (N=21) or at the top of their class (N=38) by the end of the school year. The outcome showed that three items in Harris' system (item #9 - nose; item #30 - arm; item #46 - trunk) differentiated these two extreme groups and that by confining the scoring of additional protocols of the Draw-A-Man Test to these three items alone, an improvement over Harris' 73-item scale in predicting school achievement was obtained. However, since further evidence indicated that these three key items lose their predictive potential by the end of the kindergarten, it is strongly recommended that the use of this greatly abbreviated scoring system be limited to drawings made near the start of the kindergarten year. (Author)