In pediatric cancer patients, malnutrition is commonly observed. This may represent the metabolic effect of the primary disease or it may be a consequence of multimodal therapy. This report evaluates the efficacy of using basic anthropometric measurements to predict morbidity during therapy. Twenty children with Wilms' tumor (Stage III, IV, and V) or neuroblastoma (Stage IV) diagnosed at Children's Hospital (Columbus, OH) between January 1983 and December 1985 were evaluated. When compared with the Wilms' tumor patients, the children with neuroblastoma had a significantly lower weight for age at diagnosis. At the completion of therapy, both weight-for-height and weight-for-age measurements were statistically lower in the neuroblastoma group (p less than 0.05). Significant differences were observed between the neuroblastoma and Wilms' tumor patients in the morbidity reported during therapy. Children with neuroblastoma had more frequent hospital admissions, spent a much greater proportion of their treatment time as hospital inpatients, experienced longer delays in therapy, and sustained many more complications. Each of the anthropometric indices was evaluated as a predictor of the complications observed during treatment. In the Wilms' tumor group, the patients with lower weight-for-height percentiles had an increased incidence of incomplete drug infusions, many more complications, more frequent hospital admissions, and an increase in the percentage of time spent as hospital inpatients. In the neuroblastoma group, the anthropometric measurements had no correlation with the subsequent development of complications. Nutritional staging based on anthropometric measurements recorded at diagnosis may be useful in predicting an increased risk of morbidity during therapy in children with Wilms' tumor.