In order to test the value of Hodkinson's test (HT) for the detection of dementia and other milder forms of cognitive impairment in community surveys, we have studied two separate groups of 50 individuals; one of the groups included individuals with dementia or mild cognitive impairment (MCI), and the other individuals with no cognitive disturbances. Both groups were drawn from the same rural population and subjected to HT and a neuropsychological exam for positive and differential diagnosis of cognitive impairment. The gold standard criteria for the diagnosis of dementia and MCI were the DSM-IIIR criteria and a simplified and adapted version of the criteria established for age-associated memory impairment. We found that the best cutoff of HT providing maximum sensitivity and specificity for the diagnosis of both dementia and MCI was 7. For this cutoff, sensitivity and specificity for the diagnosis of MCI were 94 and 90%, respectively; for the diagnosis of dementia sensitivity and specificity were 100 and 53%, respectively. This together with its brevity, low cost and independence of sensory impairment and cultural background makes the HT useful for large population-based screening of dementia and MCI.