Two multivariate studies examined the applicability of the theory of planned behavior in gauging students' attitudes toward mathematics, as well as the predictive power of mathematics attitudes in explaining students' grades in mathematics. Middle-school students from the United States (N = 382) and Belarus (N = 339) participated. Confirmatory factor analysis supported the viability of the theory for both samples. The analyses revealed that between 25% and 32% of the variance in mathematics grades could be explained by the theory of planned behavior components. In fact, 17% of the variation in test grades could be explained by the theory of planned behavior over and above the effects of mathematics ability test scores. Mean score differences between countries were small (ds = 0.15 to 0.27), with Belarusian students scoring more highly on attitudes and control but less highly on subjective norms and intentions. The article concludes with discussion of potential interventions and the need to expand results to different age groups and achievement domains, as well as the need for longitudinal and cross-cultural research. (Contains 2 footnotes, 5 tables, and 4 figures.)