The objective of this study was to investigate the association between specified levels of physical education (PE) at school and different parameters of cognition in boys and girls, considering potential covariates. Baseline data from a crossover nutrition intervention study were used for a secondary analysis of 247 schoolchildren (37% girls). Participants were children from the fifth and sixth grade with an "advanced" (5-6 hr) or "regular" (3 hr) level of PE a week, without random allocation. Using three computerized tasks, cognition data were collected under usual school conditions. An advanced level of PE at school was associated with better overall cognitive performance, specifically in response speed and accuracy when controlling for lunch conditions. Girls attending advanced-PE classes showed the best performance. The potential covariates explained little about cognitive variability. Given the cognitive differences between boys and girls, future research is warranted to replicate these findings by measuring motor skills.