Objective: To determine whether performance on the clock drawing test varies as a function of dementia severity and particularly whether it differentiates cognitively normal older adults from those with very mild dementia. Methods: Clock drawings were obtained during the initial clinical assessments of 75 participants in a longitudinal study of healthy aging and dementia of the Alzheimer type (15 cognitively normal, 25 with very mild dementia, 21 with mild dementia, and 14 with moderate to severe dementia, as staged by the Clinical Dementia Rating). Each clock drawing later was judged blindly and independently by two raters according to six commonly used sets of scoring criteria. Results: The same pattern of results was obtained for all six scoring methods. Inter-rater reliability was consistently high. Spearman correlations indicated that as dementia severity increased, clock drawing performance decreased. Analyses of variance revealed that although those with mild or moderate/severe dementia performed significantly worse on the clock drawing test than did individuals who had very mild dementia or who were cognitively normal, the latter two groups did not differ significantly. Sensitivity and specificity analyses revealed that all methods could distinguish between normal aging and dementia of at least mild severity. However, sensitivity for detecting very mild dementia was poor. Conclusion: Although the clock drawing test can be scored reliably and can differentiate cognitively normal older adults from those with at least mild dementia of the Alzheimer type, it does not appear to be a useful screening instrument for detecting very mild dementia.