The aim of this study is to compare the results of G and D studies that were obtained by design that formed jointly and alternatively scoring of students by more than one rater in performance assessment in terms of generalizability theory. 48 students that were chosen randomly from the third-class degree students of the Medical Faculty of Hacettepe University at 2007-2008 academic years and 3 raters constitute the study group. In this study, G and D studies that done for s x t x r and (s: r) x t designs (s:student, t:task, r:rater) are compared. As a result of study, it is observed that variances that were estimated for variables in both designs are parallel to each others. Decision studies are no such a big difference. In this way scoring certain number of students alternately is much more convenient in time, labor and economy.