One hundred and fifty unselected elderly community subjects were assessed by Mini-Mental State Examination (MMSE), Abbreviated Mental Test (AMT) and Mental Status Questionnaire (MSQ). The effects on cognitive test scores of potential confounding (non-cognitive) variables were evaluated. Sensitivities and specificities were: MMSE 80% and 98%; AMT 77% and 90%; and MSQ 70% and 89%. The MMSE identified significantly fewer false positives than the AMT and MSQ. The major effect of intelligence on cognitive test scores has previously been underestimated. Age, social class, sensitivity of hearing and history of stroke were also significantly correlated with cognitive test scores. Years of full time education and depression only affected the longer MMSE and CAMCOG. The MMSE (cut-off 20/21) can be recommended for routine screening. However, as scores are affected by variables other than cognitive function, particularly intelligence, further assessment of identified cases may fail to reveal significant functional impairment.