Two brief screening tests, the Short Portable Mental Status Questionnaire (SPMSQ) and the East Boston Memory Test (EBMT), were included in a population questionnaire administered to 3,811 persons 65 years of age and older. A detailed clinical evaluation was then administered to 467 persons (drawn from high, medium and low performers on the EBMT) to determine who was cognitively impaired and the disorders that were responsible for that cognitive impairment. The results showed that the EBMT was better at enriching the population of the poor performance group with persons who had Alzheimer's disease (AD). It had a lower refusal rate among non-proxy respondents: 2% for the EMBT versus 9% for the SPMSQ. The sensitivity and positive predictive value were also higher for the EBMT than the SPMSQ when the diagnosis of interest was AD. However, there were persons with AD in all strata of performance on both the EBMT and the SPMSQ, emphasizing the importance of selecting persons from all performance strata in multistage community studies of AD.