Background. To investigate the value of surveillance scanning for the detection of recurrences in medulloblastoma. Procedure. The charts of 95 patients with medulloblastoma were retrospectively reviewed. Information regarding the patient characteristics, treatment modalities, dates, types and results of CT and MRI studies, the frequency with which recurrences were identified on surveillance images, changes in patient management, outcome of the patients following recurrences, and survival data were analyzed. Results. Thirty-one patients had a recurrence of tumor in the central nervous system; none experienced extraneural relapses. Of all recurrences, 21 were symptomatic and 10 were discovered by surveillance scans asymptomatically. None of the patients with a recurrence survived, For all 95 patients, 5-year overall and event-free survival rates were 47.1 and 49.8%, respectively. In patients with symptomatic and asymptomatic recurrences, the mean time to recurrence since initial diagnosis, the mean duration of survival post-recurrence, and the mean duration of overall follow-up were 19.2 and 26.1 months, 3.6 and 8.0 months, and 22.8 and 34.1 months, respectively. For 95 patients, 468 surveillance and 38 symptomatic images were reviewed as 313 CTs and 193 MRIs. Rate of diagnosis of recurrence per surveillance image was 2.1% (10/468). Conclusions. In our study, Surveillance scanning brought no survival advantage since it detected a minority of recurrences. Longer survival achieved by early detection of recurrences might be a reflection of lead-time and length biases. Surveillance procedures will gain more importance as new effective therapeutic options are developed for recurrent medulloblastoma. (C) 2002 Wiley-Liss, Inc.