This paper presents the results of an experiment with 54 elementary school boys in the highlands of Peru that assessed the main effects of breakfast consumption and nutritional status (either at-risk or not-at-risk) on six tests of cognition. The subjects spent two nights at a research center, one week apart, where they were randomly assigned to either a breakfast or a no-breakfast condition on the first morning and to the opposite treatment the second morning. The same test battery was administered at 11:00 am on both occasions. It consisted of three paper-and-pencil tests (Number Discrimination, Peabody Picture Vocabulary and Raven Progressive Matrices) and three computer tests (Stimulus Discrimination, Reaction Time and Sternberg Memory Search). Whereas fasting apparently had no negative effects on the no-risk group, it delayed stimulus discrimination and memory retrieval time among the nutritionally at-risk children. Glucose levels were not associated with test performance for either group under either treatment condition.