In the course of a four-month study of interventions for behavioral disturbances in Alzheimer's disease (AD) patients, the following assessment instruments were examined for validity: the clinical global impression of change (CGIC), Cohen-Mansfield agitation inventory (CMAI); CERAD behavioral rating scale for dementia (BRSD), revised memory and behavioral problems checklist (RMBPC) and the agitated behavior in dementia scale (ABID). The four specific behavioral/agitation scales had excellent cross-sectional and longitudinal correlations with each other, suggesting high validity, but changes as indicated by CGIC scores did not correlate well with change scores on the other instruments. We conclude that specific behavioral instruments are preferable to the more general CGIC for detecting and quantifying behavioral disturbances in AD patients.