Successful surgical management of a neoplastic or nonneoplastic seizure focus in close proximity to or within eloquent brain areas relies on precise delineation of the relationship between the lesion and functional brain areas. The aim of this series was to validate the usefulness and test the efficacy of noninvasive presurgical PET mapping of eloquent brain areas to predict surgical morbidity and outcome in children with seizures. To identify eloquent brain areas in 15 children (6 female and 9 male; mean age 11 years) with epileptogenic lesions PET images of regional cerebral blood flow were performed following the administration of [15O]water during motor, visual, articulation, and receptive language tasks. These images with coregistered magnetic resonance (MR) images were then used to delineate the anatomic relationship of a seizure focus to eloquent brain areas. Additional PET images using [18F]fluoro-2-deoxy-D-glucose (FDG) and [11C]methionine (CMET) were acquired to help localize the seizure focus, as well as characterize the lesion. Patient surgical management decisions were based on PET mapping in combination with coregistered MR images, PET-FDG findings, and the anatomic characteristics of the lesion. At follow-up 1–26 months after surgery, all patients that underwent temporal lobectomy (9 patients) and extratemporal resection (4 patients) for a neoplastic or nonneoplastic seizure focus are seizure-free with minimal postoperative morbidity. Of prime importance, no child sustained a postoperative speech or language deficit. PET imaging was also well tolerated without procedural complications. Based on PET mapping, a nonoperative approach was used for 2 children and a biopsy only was used in one child. When cortical injury involved prenatally determined eloquent cortex, PET demonstrated reorganization of language areas to new adjacent areas or even to the contralateral hemisphere. Integration of anatomical and functional data enhanced the surgical safety, defined optimal surgical approach, delineated the seizure focus from eloquent brain areas, facilitated maximum resection and optimized the timing of surgery, thereby minimizing surgical morbidity while maximizing surgical goals. PET measurements of FDG and CMET uptake were also helpful in localizing the seizure focus and grading the tumors. PET used for brain mapping in children provides the surgeon with strategic preoperative information not readily attainable with traditional invasive Wada testing or intraoperative cortical stimulation. PET mapping may also improve the outcome of extratemporal resections by allowing aggressive seizure focus resection. In addition, serial brain maps may optimize timing for surgical intervention by demonstrating reorganization of eloquent cortex often seen in younger children after cortical injury. Our results suggest that noninvasive presurgical brain mapping has the potential to reduce risk and improve neurologic outcome.