Purpose: To study the role of magnetoencephalography (MEG) in the surgical evaluation of children with recurrent seizures after epilepsy surgery. Methods: We studied 17 children with recurrent seizures after epilepsy surgery using interictal and ictal scalp EEG, intracranial video EEG (IVEEG), MRI, and MEG. We analyzed the location and distribution of MEG spike sources (MEGSSs) and the relationship of MEGSSs to the margins of previous resections and surgical outcome. Results: Clustered MEGSSs occurred at the margins of previous resections within two contiguous gyri in 10 patients (group A), extended spatially from a margin by ≤3 cm in three patients (group B), and were remote from a resection margin by >3 cm in six patients (group C). Two patients had concomitant group A and C clusters. Thirteen patients underwent second surgeries. IVEEG was used in four patients. Six of seven patients with group A MEGSS clusters did not require IVEEG for second surgeries. Follow-up periods ranged from 0.6 to 4.3 years (mean: 2.6 years). Eleven children, including eight who became seizure-free, achieved Engel class I or II. Conclusion: Our data demonstrate the utility of MEG for evaluating patients with recurrent seizures after epilepsy surgery. Specific MEGSS cluster patterns delineate epileptogenic zones. Removing cluster regions adjacent to the margins of previous resections, in addition to removing recurrent lesions, achieves favorable surgical outcome. Cluster location and extent identify which patients require IVEEG, potentially eliminating IVEEG for some. Patients with remotely located clusters require IVEEG for accurate assessment and localization of the entire epileptogenic zone. © 2007 International League Against Epilepsy.