CTV margins for pHGG are derived from adult HGG neuroimaging and pathology data. We developed a novel algorithm to estimate the margin required to cover the volume at risk delineated by anatomic and metabolic imaging. This method may help clinicians to objectively define anisotropic CTV margins. Twenty-six pediatric patients with HGG experienced local tumor progression after chemoradiotherapy. Progression was central, infield, and marginal in 50%, 37.5%, and 12.5% of cases, respectively. Anatomic epicenter, RANO failure type (clinical symptoms, >25% increased in T1Gd, etc.) and steroid use at progression (33%) were recorded. Tumor volumes and surgical cavities were delineated on MRI (T1+Gd, T2, FLAIR) and methionine-PET (MET-PET) at diagnosis (DTV) and failure (FTV). The distance between DTV and FTV surfaces was measured using a novel algorithm. The net progression vector (NPV), defined as the vector length approximating 80% of vectors required to cover the FTV, was calculated. Component progression vectors (CPV) were deconstructed from NPV to their x, y, and z components to estimate anisotropy in the pattern of progression (x = medial-lateral (ML), y = anterior-posterior (AP), z = craniocaudal (CC)). Comparisons of continuous and count data across disease and treatment factors were summarized using either ANOVA or the Chi-Square test. The NPV defined by T2 FLAIR and T1+Gd was 1.9 cm and 1.8 cm, respectively. NPV measurements were not significantly different according to extent of resection. NPV margin estimates differed substantially across RANO failure type. Patients exceeding RANO measured failure thresholds had increased NPV margin estimates. CPV measured 0.9 cm, 0.8 cm, and 1.0 cm in the x, y, and z directions, respectively, suggesting moderate anisotropy for all tumor locations. Midbrain, frontal, and temporal lobe tumors exhibited anisotropic extension. Frontal and midbrain tumors exhibited reduced CC extension, while frontal-temporal lobe tumors exhibited substantial AP extension. Steroid requirements at failure selected for patients with increased FLAIR defined NPV (2.2 cm vs. 1.9 cm, P = 0.15) but not the T1+Gd defined NPV. Patients with marginal failure had mildly increased NPV relative to central/infield failures (2.4 vs. 1.8 cm, P = 0.2). The MET-PET delineated NPV showed significant agreement with NPV derived from T1+Gd and T2 FLAIR imaging. The MET-PET delineated DTV overlapped the FTV in 90% of patients. NPV may be a useful method to define CTV margins in pHGG, especially for target volumes with complex geometry. Use of steroids at the time of tumor progression, RANO failure type and tumor location impacted the NPV required to cover the FTV. CPV sampling is a useful way of understanding the anisotropy of tumor extension.