The present study was designed to examine both concurrent and predictive associations between scores on a measure of executive function (EF) skills, the Contingency Naming Test (CNT), during the early school-age years. A secondary aim of the study was to examine the association between EF skills and mathematics performance. We administered tests of mathematics ability, and the CNT, to 178 children at ages 6 to 7, 8 to 9, and 10 to 11 years. From the CNT we obtained measures of response fluency/efficiency, working memory, and inhibition. The results demonstrate main effects of age on all CNT measures of EF, as anticipated, and inconsistent main effects of gender or mathematics learning disability status. Rates of improvement in EF varied as a function of the working memory demands present during a given task. There were differences in concurrent and predictive correlations for different CNT performance measures. EF scores obtained during the first assessment were as strongly associated with each other as they were with EF scores obtained four years later, suggesting a moderately stable source of individual differences on cognitive performance. EF scores at age 6 to 7 years were associated with concurrent and later mathematics scores, and most of these correlations were stronger than the significant associations found between response fluency on a baseline task (with no working memory demand) and mathematics performance. These findings have implications for the stability of EF skills during the school-age years, and the role of EF in early and later elementary school mathematics performance.