Background: Diagnosis of mild cognitive impairment (MCI) and its subtypes requires a detailed diagnostic assessment and so may be missed at routine primary-care appointments, since the ordinary dementia screening tests lack sensitivity owing to ceiling effects, especially in highly educated subjects.

Methods: The study was undertaken using a cross-sectional assessment of 112 elderly subjects (mean age 67.96 ± 5.77 years, and mean education level of 12.8 ± 5.7 years) with a semi-structured interview and a neuropsychological battery.

Results: MCI patients did not differ from controls on total MMSE scores (p = 0.212). Nevertheless, MCI patients showed worse performance than controls on the verbal memory task (p = 0.012), and “drawing a pentagon” (p = 0.03). Amnestic MCI patients performed worse only on the “three-word recall” task (p = 0.013); non-amnestic MCI patients performed worse on the “three-stage command” task (p = 0.001); and multiple-domain MCI patients performed worse on the “drawing a pentagon” task (p = 0.001) and had a trend toward performing poorly on the ‘three word recall’ task (p = 0.06).

Conclusion: The analysis of MMSE subtest scores, in addition to MMSE total scores, may increase the sensitivity of the MMSE test in screening for MCI and its subtypes.