The primary objective of this cross-sectional study was to test the hypothesis that the plasma concentration of insulin-like growth factor I (IGF-I) correlates with the risk of in-hospital morbidity among metabolically stable elderly. The secondary objective was to determine whether IGF-I correlates with other putative indicators of protein-energy nutritional status.To meet these objectives, 110 randomly selected admissions to a Geriatric Rehabilitation Unit (GRU) of a Veterans Administration hospital were studied. The average age of the study patients was 76 years; 98% were male, and 78% were white.At admission, each patient completed a comprehensive medical, functional, neuropsychological, and nutritional assessment including the attainment of a plasma IGF-I determination. While in the hospital, each subject was monitored daily for development of complications.Subsequent to GRU admission, 44 patients (40%) experienced at least one complication; 36 patients (33%) experienced an infectious complication, 8 patients (7%) experienced a life-threatening complication, and 5 patients (5%) experienced a life-threatening infectious complication. There was no correlation between IGF-I and development of a non-life-threatening complication (”any complication”or“any infectious complication”). However, IGF-I was a strong predictor of“life-threatening”and“life-threatening infectious”complications. The logistic regression model incorporating the independent variable IGF-I differentiated the patients who would develop a“life-threatening”complication from those who would not with a sensitivity of 75%, a specificity of 76%, and an overall predictive accuracy of 76%. IGF-I was highly correlated with admission serum albumin, transferrin, and cholesterol, triceps skinfold thickness, body weight expressed as a percent of ideal, and body mass index. There was no correlation between IGF-I and the Katz Index of Activities of Daily Living score, age, or anergy status.IGF-I is a strong predictor of“life-threatening”and“life-threatening infectious”complications and may be a clinically useful marker for protein-energy undernutrition among metabolically stable hospitalized elderly patients.