It is with increasing importance that mathematics education research considers the role of noncognitive motivation variables alongside elements of the classroom context in investigations of student mathematics performance. This study uses a hierarchical linear modeling framework to predict mathematics achievement from three classroom variables, project-based learning, group collaboration, and student-driven curriculum, and two noncognitive factors, mathematics anxiety and mathematics self-concept, utilizing data from the 2012 Programme for International Student Assessment (PISA) United States sample. Findings suggest that mathematics classroom contexts that are student-driven and integrate project-based learning positively impact mathematics achievement, and that both mathematics anxiety and mathematics self-concept contribute significantly towards explaining variation in mathematics achievement after accounting for gender, race, socioeconomic status, truancy, and school-level poverty.