Humans are an exceptionally intelligent species, and the selective pressures which may have shaped these advanced cognitive powers are therefore of interest. This study investigates the fitness consequences of pre-reproductive school performance in a Swedish population-based cohort of 5244 males and 4863 females born 1915-1929. School performance was measured at around age 10 using three variables: mean school marks, being promoted/held back in school, and recognised learning difficulties. Our primary outcomes were probability of ever marrying, total number of children and total number of grandchildren. In males (but not females), poorer school performance predicted fewer children and grandchildren. This was primarily mediated via probability of marriage; mortality and fertility within marriage were not important mediating pathways. The effect of school performance upon marriage in males was independent of early-life social and biological characteristics, including birth weight for gestational age, preterm birth, family composition, and family socioeconomic position. The effect of school performance upon the probability of marriage in males was, however, largely mediated by adult socioeconomic position. This suggests that in general sexual selection for cognitive abilities per se did not play a major role in either males or females in this cohort. Adult socioeconomic position did not, however, fully explain the marriage disadvantage in males or (at marginal significance) females with particularly poor school performance. We conclude that school performance can affect long-term reproductive success. In this population, however, the effect is confined to males and is largely mediated by the increased probability of marriage which comes with their greater socioeconomic success. © 2010 Elsevier Inc.