To characterize qualitative Clock Drawing Test (CDT) error profiles across dementia etiologies and mild cognitive impairment (MCI), and to propose a clinically interpretable six-class framework.

In a hospital-based study in Kolkata, India, consecutive adults with cognitive impairment completed a free-drawn “ten past ten” CDT. Errors were coded using classical qualitative categories and clock components (face, numbers, hands), then collapsed into six classes: conceptual, stimulus-bound/perseveration, spatial, planning, number-related, and graphic-conceptual. For nonexclusive domains, omnibus Pearson χ² tests summarized error distributions across diagnoses; Cramér’s V quantified effect size.

Participants included Alzheimer’s disease (AD; n = 36), vascular dementia (VaD; n = 16), behavioral variant frontotemporal dementia (bvFTD; n = 9), MCI (n = 19), and other conditions (n = 22). Although 50.0% drew a normal clock face, only 11.8% achieved perfect numbering and 12.7% set the hands correctly. Conceptual errors were most frequent (70.6%), followed by spatial errors (47.1%); neglect and counterclockwise numbering were rare (3.9% each). Error distributions differed by diagnosis for face, numbers, and hands (all p < 0.001; V = 0.4–0.5). The six-class scheme retained a significant distributional association with diagnosis (χ² = 43.365, p = 0.002; V = 0.3): bvFTD showed prominent conceptual and graphic-conceptual failures, AD combined conceptual and spatial errors, MCI emphasized spatial and number-related errors, and VaD was heterogeneous.

Qualitative CDT profiles vary meaningfully across cognitive disorders. This concise six-class framework captures clinically salient patterns, especially in severely degraded drawings, and may complement brief memory screening and digital CDT metrics.