The goal of this work is to determine whether we can identify the children with acute lymphoblastic leukemia (ALL) who will have cognitive and academic problems using markers of central nervous system (CNS) injury obtained during treatment. Although declines in cognitive and academic function have been linked to CNS treatment in many studies, some ALL survivors exhibit significant declines while others do not. What accounts for this variability? Our recent findings suggest that changes in CSF phospholipid composition during treatment are related to subsequent cognitive and academic function. Based on our preliminary studies and available literature, the investigators believe that CNS injury has both direct and indirect effects on cognitive and academic function and that one indirect effect is a result of the impact of CNS injury on CNS processing speed. Thus, they will examine relationships between CSF markers of CNS injury, processing speed, and cognitive and academic abilities using latent variable and structural equation models. This prospective longitudinal study will follow 90 children with ALL who are age 3 to 13 years of age at diagnosis and treated according to Pediatric Oncology Group Protocol with triple intrathecal chemotherapy and intermediate dose systemic methotrexate. CSF concentration of 3 phospholipids (sphingomyelin, phosphatidylcholine, lysophosphaditylcholine) and cardiolipin will be used as markers of CNS injury. Cognitive and academic abilities will be evaluated at diagnosis and 1, 2,and 3 years after diagnosis. The investigators will use growth curve analysis to estimate each child's initial status and rate of change on cognitive and academic measures. They hypothesize that there will be (1) A significant negative relationship between CSF markers and performance on cognitive and academic measures; (2) A significant negative relationship between CSF markers and processing speed; and (3) A significant positive relationship between processing speed and cognitive and academic performance. Variables that may be related to cognitive outcomes will also be included in measurement models. These include the child's age at diagnosis or evaluation, sex, baseline abilities, school attendance and educational services and socioeconomic status. A sibling comparison group will be followed to allow estimation of loss of abilities associated with ALL and its treatment and identify cognitive abilities that may be particularly vulnerable. If the study can identify children at risk for cognitive and academic problems, those most in need of assistance can be provided with educational services and evaluation interventions.