Mental health problems during childhood are associated with lowered educational attainment in adulthood. However, it is not clear if these associations hold when controlling for unobserved features of the family environment and if they depend on the socioeconomic status (SES) of parents. We use the Panel Study of Income Dynamics (PSID) Child Development and Transition into Adulthood Supplements (CDS; TAS) to examine these questions. Using linear and logistic regression, we isolate within-family variability in mental health problems among full sibling pairs (n = 958 individuals in 479 pairs). Associations depend on the measure used. Parental reports of problem behaviors and diagnosed problems have the most consistent negative associations with educational attainment (for example, a hyperactivity diagnosis is associated with 0.74 fewer years of schooling). Retrospective self-reports of diagnoses other than depression or anxiety also have a negative association (0.96 fewer years of schooling). But self-reports of depressive symptoms and emotional or psychological well-being during late childhood and adolescence have no significant associations with educational attainment. In addition, there is no significant moderation of these associations by SES.