OBJECTIVE The primary objective of this study was to determine the efficacy of the clock drawing test to predict the presence of very mild Alzheimer's disease (AD). A secondary objective was to identify elements of clock drawing that were most useful in differentiating cognitively intact older adults from those with mild Alzheimer's disease. METHODS Cohort based comparison of retrospective data. METHODS Academic research center. METHODS Clock drawings from 41 outpatient cases of mild AD with Mini-Mental State Exam scores of 24 or higher and 39 age- and education-matched older adults were scored. METHODS Clock drawings were blindly and independently scored by two raters using the Clock Drawing Interpretation Scale and the scoring system reported by Rouleau et al. Predictive values for positive and negative tests were calculated using cut-off scores for total score and component subscores from each of these two systems. RESULTS Two or more errors in the depiction of the clock hands on the Clock Drawing Interpretation Scale had a positive predictive value for AD of 100% and a negative predictive value of 51%. A score of 2 or less on the 4-point hand-placement component of the Rouleau et al. scoring system provided a positive predictive value for AD of 94% and was associated with a negative predictive value of 62%. CONCLUSIONS An individual who commits two errors or more in drawing the clock hands deserves further investigation for a possible dementia. Normal hand placement on the clock drawing test does not exclude AD. However, when prevalence rates of dementia in community-dwelling older adults are considered, these results argue that normal clock hand placement indicates that dementia is unlikely.