Studies of an association between school performance (enrollment, attendance, achievement, classroom behavior, and school drop-out) and nutritional status are discussed and the evidence of an effect of school meals evaluated. Many studies have shown associations between hunger, poor dietary intakes, stunting, underweight, and poor school performance after controlling for socioeconomic conditions. However, it remains possible that unmeasured socioeconomic variables could explain the relationship. Rigorous short-term studies of missing breakfast have generally found detrimental effects on children's cognition whereas studies of providing breakfast have shown benefits particularly in malnourished children. Classroom conditions may modify the effects of breakfast on behavior. There are extremely few longer term studies of the effects of giving school meals. Nearly all involved breakfast and very few had randomized controlled designs. Studies comparing participants with non-participants or comparing matched schools have found benefits (but bias due to self-selection) of receiving breakfast; inadequate matching of schools also remains possible. One longer term randomized controlled trial found benefits associated with attendance and arithmetic performance. In conclusion, most studies of giving breakfast have found benefits to school performance but many had serious design problems, were short-term, and were not conducted in the poorest countries. In order to advise policy makers correctly, there is an urgent need to run long-term randomized controlled trials of giving school meals in poor countries and to determine the effects of age and nutrition status of the children, the quality of the school, and the timing of the meal. The special needs of orphans should also be considered.