INTRODUCTION: Central nervous system prophylaxis of childhood acute lymphoblastic leukemia has dropped rates of relapses but has been associated with neurotoxicity and imaging abnormalities. Predictors of neurotoxicity are lacking, because of inconsistency between clinical symptoms and imaging. Some have suggested that cerebrospinal fluid myelin basic protein (MBP) levels to be of potential interest. A retrospective analysis of MBP levels in correlation with clinical and radiologic data is presented. MATERIALS AND METHODS: MBP levels obtained at the time of intrathecals, charts, and neuroradiology reports were retrospectively analyzed. Academic achievement data were obtained from phone contacts with patients and families. RESULTS: We retrieved 1248 dosages of MBP in 83 patients, 381 neurologic examinations in 34 patients and 69 neuroradiologic investigations in 27 patients. Fifty-two patients had abnormal MBP levels. Radiologic anomalies were present in 47% of those investigated, 14% of them having school difficulties. Proportions of patients with school difficulties in the groups with abnormal MBP levels but no radiologic anomalies or with no radiologic investigations were 0% and 3%, respectively, which was lower than in the group of patients with normal MBP levels (100%, 22%, and 5%, respectively). DISCUSSION: Notwithstanding the retrospective character of our study, we conclude that there is limited usefulness of systematic dosage of MBP as indicator of treatment-induced neurotoxicity in acute lymphoblastic leukemia patients.