The Mini-Mental State Examination (MMSE) is commonly used as a screening test for dementia, yet MMSE scores above the cut-off for dementia (24–30) are widely thought to have limited utility, particularly in older persons. The study investigates whether scores within this range can be indicative of pre-symptomatic levels of cognitive impairment. Ninety-six community-dwelling older persons aged 62–89 years (mean = 75.2 years), who had obtained MMSE scores between 25 and 30, were tested on the computer-based Subtle Cognitive Impairment Test (SCIT). Compared with individuals who obtained a perfect score of 30 on the MMSE, individuals with scores of 28–29 made more errors on the SCIT, whereas those with scores of 25–27 on the MMSE made the most errors on the SCIT (F(2,94) = 9.84, p < 0.01). Individuals who made errors in the language (r(94) = −0.47, p < 0.01), attention (r(94) = 0.24, p < 0.05) and visual construction (r(94) = −0.27, p < 0.01) subtests of the MMSE were more likely to display impaired SCIT performance. The pattern of performance on the SCIT varied in a systematic way, depending on the MMSE subtest in which the errors were made, raising the possibility that there may be different subtypes of subtle cognitive impairment within the ostensibly normal population of older persons.