Objective: Clock Drawing Test (CDT) has been widely used in dementia screening as a practical and valid tool assessing various neuropsychological domains. Many scoring systems requiring different instructions and methodologies were proposed for CDT and there are several studies comparing the diagnostic accuracy of these methods. However, there is still no clear agreement on which one is the best for clinical and research settings. The present study aimed to investigate the psychometric properties of the Turkish versions of the three most commonly used CDTs (Shulman's, Sunderland's and Watson's CDTs) and to compare their diagnostic accuracy with each other and Mini Mental State Examination (MMSE).Methods: The study sample consisted of elderly patients with dementia (n=65) and healthy controls (n=112) who were admitted to the Department of Geriatric Psychiatry and Geriatrics of a University hospital. The subjects were administered MMSE and CDT. In CDTs, the subjects were asked to place the numbers and the hands of a clock in a pre-drawn circle showing the time setting for the clock as "10 after 11". CDTs were scored according to three different CDT scoring methods (Shulman, Sunderland and Watson) by two blinded raters. Discriminant and concurrent validity and interrater reliability were assessed. Receiver operating characteristic (ROC) analysis was also performed to calculate cut-off scores and sensitivity/specificity values. Areas under the curve (AUCs) of CDTs were compared. Results: The mean scores of all CDTs were significantly different in dementia patients compared to the controls. CDT scores were significantly correlated with MMSE scores. All CDTs had high sensitivity values, while Watson's CDT had lower specificity. AUC value for MMSE was higher than that for CDTs and AUC value for Watson's CDT was lower than the values for the other two CDTs. When Shulman/Sunderland CDTs were used together with MMSE, the sensitivity increased up to 100%.Conclusion: The Turkish versions of all three CDTs can be used in dementia screening. Shulman's or Sunderland's CDT may be preferred to Watson's CDT, since they have higher specificity and diagnostic accuracy. © Archives of Neuropsychiatry.