Many studies have now confirmed the association between inheritance of the epsilon 4 allele of the apolipoprotein E (APOE) gene and Alzheimer disease (AD). However, although the medical community holds the near-unanimous opinion that APOE genotyping should not be used for prediction in asymptomatic individuals, controversy remains about whether it should be used for diagnosis in patients who show signs of dementia. We assessed critically the recent clinical studies, on the basis of four criteria recommended to ensure safety and effectiveness of genetic tests. We also developed a formal framework for evaluating the usefulness of APOE genotyping using decision-theoretic principles. We conclude that neither the presence nor absence of an epsilon 4 allele provides diagnostic certainty, and the proper interpretation of either result in heterogeneous populations requires further investigation. The appropriate role of APOE genotyping among elements of a traditional assessment for AD has not been determined. Whether APOE genotyping provides sufficient information to change patient management decisions has not been determined. APOE genotyping presents foreseeable, significant psychosocial consequences for family members that must be weighed against any psychosocial benefits. Therefore, the diagnostic use of APOE genotyping outside research settings is premature until such testing is shown to be of practical value.