Objective: To compare the merits of regional ultrasonography, anthropometry, and measurement of visceral protein concentrations for determining muscle protein mass and nutritional status. Design: Prospective follow-up study. Subjects: Nineteen children (median age 4.7 y, range, 1.4 ±14.1 y) with malignant solid tumors were examined at diagnosis, monthly during preoperative chemotherapy, and 1±2 months after operation; 19 healthy age-and sex-matched children served as controls. Interventions: We measured the thickness of the quadriceps muscle by ultrasonography (muscle index), and body weight, mid-arm circumference, triceps skinfold thickness, and serum concentrations of albumin, prealbumin, and transferrin. Results: At diagnosis, the muscle index was smaller in the patients (mean 5.6 cm 2 am 2 , 95% CI 4.5 ± 6.7 cm 2 am 2 ) than in the controls (8.5 cm 2 am 2 , 95% CI 7.7±9.3 cm 2 am 2 ) (P 0.0007), whereas the anthropometric measurements did not differ. The sensitivity, speci®city, and predictive value of anthropometry and measurements of visceral proteins were poor in detecting reduced muscle protein reserves when compared with the muscle index as the baseline standard. During the anticancer therapy the mean concentrations of serum albumin and prealbumin increased. Conclusions: In children with cancer, nutritional status should be assessed with methods more sensitive than anthropometry and visceral protein concentration measurements. We advocate regional ultrasonography for the evaluation of nutrition in these patients. Sponsorship: This work was supported by the Foundation for Pediatric Research, Helsinki, Finland.