Background. The performance of the CES-D in a sample of elderly community residents was assessed. The influence of dementia on test performance and the necessity for the use of four factor scores instead of a single summary score of the CES-D were studied.

Method. Two hundred and eighty-seven subjects out of the general population aged 60–99 years were personally interviewed with standardized diagnostic tools and completed the CES-D. Best-estimate diagnoses served as ‘gold standards’ for receiver operating characteristics (ROC) analysis.

Results. The CES-D discriminated well between depressive and non-depressive subjects. Exclusion of demented subjects from the sample did not markedly increase test performance. Current depressive illness and dementia led to high scores on the CES-D. Unlike the factors ‘depressive affect’, ‘somatic/vegetative complaints’, and ‘interpersonal relations’, the factor ‘positive affect’ of the CES-D discriminated well between demented and non-demented participants.

Conclusions. The CES-D is a valid instrument for screening for depression in a community sample of elderly subjects. Its use can be recommended even if the presence of dementia is likely. The use of factor scores of the CES-D does not substantially contribute to an improvement of overall test performance, but, nevertheless, allows a more detailed insight and better interpretation of test results.