Providing two or more versions of multiple-choice exams has long been a popular strategy for reducing the opportunity for students to engage in academic dishonesty. While the results of studies comparing exam scores under different question-order conditions have been inconclusive, the potential importance of contextual cues to aid student recall and recognition has been a common theme. The current study compares the effect of forward- and random-ordered exam questions on grades in two courses, introductory psychology and principles of learning, with different degrees of interrelatedness among the topics covered. Context cues would presumably be more important in courses with relatively independent chapter topics. No significant differences were noted, highlighting the need for a new framework for understanding factors that affect the reliability of different test versions.