The staging of dementia is ideally based on both an examination of the patient and a history taken from an informant. However, in some circumstances, only an informant history is possible. The aim of this study was to assess the validity of the Clinical Dementia Rating (CDR) when the rating is based solely on informant data. The CDR was used in a study of 360 persons aged 78 or older who were participants in a community survey, the Sydney Older Persons Study. The CDR was completed in two ways: (1) a social scientist made the ratings based on an informant interview; and (2) a physician made the ratings after an examination of the subject. All CDRs were scored in the conventional way, as well as by the revised method proposed by Gelb and St. Laurent (Alzheimer Dis Assoc Disord 1993;4:202-11). The informant CDR showed moderate agreement with the clinician CDR, showing that it would be a valid substitute in situations in which the subject could not be examined. The revised scoring method was slightly easier to implement than the conventional method.