The standard follow-up care for children with cerebellar astrocytomas includes regular surveillance imaging of the brain with computerized tomography or magnetic resonance. The purpose of surveillance imaging is to detect asymptomatic tumor recurrence at an early stage and permit safer reoperation. The authors evaluated the effectiveness of an intensive surveillance program for cerebellar astrocytoma and tested different models of surveillance frequency and duration to arrive at a specific recommended program. Review of the records of 93 children with typical cerebellar astrocytomas who received follow-up care between 1975 and 1993 was performed. Immediate postoperative and surveillance images were classified as showing definite equivocal, or no tumor based on the radiology report at the time the image was obtained. Various surveillance models were then tested for their predictive value for detecting tumor recurrence. Seventeen (18%) of the 93 children had tumor recurrence or progression. Eleven of these tumors were asymptomatic and detected only by surveillance image. Tumor recurred in only one patient with a total resection, whereas tumor progression occurred in five of 21 patients with equivocal postoperative images and in 11 of 14 patients with residual tumor. A model in which patients with possible or definite residual tumor after surgery undergo surveillance at 12, 18, 30, 42, and 66 months, and later have one additional image, yielded optimum predictive value for recurrence and/or progression with the fewest images. Patients with tumor recurrence were satisfactorily treated, and only one patient died. Children with totally resected cerebellar astrocytomas do not appear to benefit from routine surveillance, because the likelihood of recurrence is small. Surveillance is of benefit in those who may have subtotal resection based on the immediate postoperative imaging.