This paper reviews the issues involved in using statistical data on multiple choice examination results as evidence of cheating among college student test-takers. Recent studies have demonstrated the large extent of academic dishonesty among American college students. Seeking to curb this trend, college faculty have been turning to statistical methodologies to detect cheating on multiple choice examinations. The paper maintains that no mechanistic detection method currently available can provide reliable evidence of cheating. Statistical evidence alone should not be used to accuse individuals of cheating, it is argued, since it cannot conclusively prove that cheating took place. The paper concludes by asserting that faculty and administration must work together to change the culture surrounding academic dishonesty from discipline to development, from prosecution to prevention. (Contains 34 references.) (MDM)