BACKGROUND The subjective global assessment (SGA) is a simple, sensitive tool used to identify nutritional risk. It is widely used in the adult population, but there is little evidence on its effectiveness in children with cancer. OBJECTIVE This cross-sectional study was undertaken to demonstrate significant correlation between a simplified version of the Patient-Generated SGA (PG-SGA) and anthropometric assessment to identify nutritional status in children recently diagnosed with cancer. METHODS The nutritional status of 70 pediatric cancer patients was assessed with the PG-SGA and anthropometric measurements. The relation between the assessments was tested with ANOVA, independent samples t-test, Kappa statistic, and non-parametric Spearman and Kendall correlation coefficient. The PG-SGA divided the patients into four groups: well nourished, mildly, moderately and severely malnourished. RESULTS The prevalence of malnutrition according to the PG-SGA was 21.4%. The correlations (r ≥ 0.300, p < 0.001) and the concordance (k ≥ 0.327, p < 0.001) between the PG-SGA and anthropometric indicators were moderate and significant. CONCLUSIONS The results indicate that the PG-SGA is a valid tool for assessing nutritional status in hospitalized children recently diagnosed with cancer. It is important to emphasize that the subjective assessment does not detect growth retardation, overweight or obesity.