This paper examines the usefulness of various anthropometric classification systems of nutritional status in prognosticating the subsequent risk of mortality among 2019 children aged 13 to 23 months residing in a rural area of Bangladesh. The indices investigated included: weight-for-age; weight-for-height; height-for-age; arm circumference-for-age; arm circumference-for-height; weight quotient; and height quotient. Cross-sectional anthropometry was conducted during October 1975 to January 1976 and the mortality experience of the study children was followed prospectively over 24 months. Results indicated that severely malnourished children, according to all indices, experienced substantially higher mortality risk. Normal, mild, and moderately malnourished children all experienced the same risk. All indices were found to discriminate mortality risk; weight/age and arm circumference/age were strongest and weight/height weakest. For each index, a threshold level was noted below which mortality risk climbed sharply. The discriminating power of anthropometry was enhanced when maternal weight, maternal height, or housing size were included.