Extreme body mass index (BMI) values (i.e., above the 97th and below the 3rd percentiles) are inaccurately represented on the Centers for Disease Control and Prevention’s growth curves, which may limit the utility of BMI percentile and BMI z-score for capturing changes in clinical outcomes for patients at extreme weights. Modeling child obesity severity based upon the percentage of BMI in excess of the 95th percentile (BMI95pct) has been proposed as an improved metric to better capture variability in weight at extreme ends of growth curves, which may improve our understanding of relationships between weight status and changes in clinical outcomes. However, few studies have evaluated whether the use of BMI95pct would refine our understanding of differences in clinical psychosocial constructs compared to previous methods for categorization. This cross-sectional study evaluated child obesity severity based on BMI95pct to examine potential group differences in a validated, obesity-specific measure of Health-Related Quality of Life (HRQoL). Four hundred and sixty-five children with obesity completed Sizing Me Up, a self-report measure of HRQoL. Children were classified into categories based on BMI95pct (i.e., class I: ≥100% and