Reliable quantitative ratings of chronic medical illness burden have proved to be difficult in geropsychiatric practice and research. Thus, the purpose of the study was to demonstrate the feasibility and reliability of a modified version of the Cumulative Illness Rating Scale (CIRS; Linn et al., 1968) in providing quantitative ratings of chronic illness burden. The modified CIRS was operationalized with a manual of guidelines geared toward the geriatric patient and for clarity was designated the CIRS(G). A total of 141 elderly outpatient subjects (two medical clinic groups of 20 each, 45 recurrent depressed subjects, 21 spousally bereaved subjects, and 35 healthy controls) received comprehensive physical examinations, reviews of symptoms, and laboratory testing. These data were then used by nurse practitioners, physician's assistants, and geriatric psychiatrists to compute CIRS(G) ratings of chronic illness burden. As hypothesized, analysis of variance demonstrated significant differences among groups with respect to total medical illness burden, which was highest among medical clinic patients and lowest in control subjects. Good interrater reliability (i.e., intraclass correlations of 0.78 and 0.88 in a subsample of 10 outpatients and a separate group of 10 inpatients, respectively) was achieved for CIRS(G) total scores. Among medical clinic patients, a significant correlation was found, as expected, between CIRS(G) chronic illness burden and capability as quantified by the Older Americans Activities of Daily Living Scale; and between CIRS(G) scores and physicians' global estimates of medical burden. Finally, with repeated measures of illness burden approximately 1 year from symptom baseline, significant rises were detected, as expected. The current data suggest that the CIRS(G) can be successfully applied in medically and psychiatrically impaired elderly subjects, with good interrater reliability and face validity (credibility).