The present study compares the sensitivity and specificity of the short and full forms of the Cambridge Examination for Mental Disorders of the Elderly (CAMDEX) interview in diagnosing dementia. We tested 73 subjects meeting DSM-HIR criteria for dementia and 61 matched controls. The short version was applied 3 months after the full one to guarantee a relative stability of the tested functions. Referred to an independent clinical rating made at the beginning of the study, the levels of sensitivity and specificity were not significantly different in the two forms and fully comparable with those of the original full English version. Moreover, the scores on analogous sections of the two versions were highly correlated in the demented and control groups. These findings support the hypothesis that the short form of the CAMDEX maintains the psychometric properties of the full one, and consequently can be used in diagnostic routines for a variety of clinical and research purposes.