Two experiments used two 2 x 3 x 2 (instructors x test forms x class times) designs to compare three aspects of student performance (mean final examination scores, homogeneity of variance, and retake frequencies) on three test forms. The test items on the three forms were arranged in three ways: (1) forward--items paralleled the presentation of material in class; (2) backward--items were in an opposite order; and (3) random order. The 336 undergraduates were taught using either Contingency Managed Lecture (CML) or the Keller Personalized System of Instruction (PSI). Analyses indicated that: (1) item order alone did not significantly influence student performance in either experiment; (2) test form and class time significantly interacted for CML; (3) none of the item orders affected retake rates in either experiment; (4) analysis of variance of all twelve cells for each experiment revealed heterogeneity only for CML and only by collapsing across the time variable; and (5) the CML experiment also revealed an instructor x test form interaction. It was concluded that there was no evidence that test order disrupted performance, in spite of the fact that there was tendency for random order to produce the best results. (Author/GDC)