The primary purpose of this quasi-experimental mixed-methods study was to examine whether a blended learning approach was more effective than traditional face-to-face instruction in improving the mathematical achievement and attitudes of preservice elementary teachers. The secondary purpose was to determine what specific features of blended learning were deemed most helpful to prospective teachers' learning of mathematics. Sixty-three prospective teachers participated in a foundational mathematics course in either a face-to-face (F2F) traditional or blended learning format. At pre- and post-test, participants completed a researcher-developed mathematics achievement test and the Dutton's Attitude Scale (Dutton in ESJ 68:259--264, 1968) . Blended learning participants also completed a post-course researcher-developed questionnaire on their perceptions of blended learning. Results indicated that participants who were enrolled in the blended learning format had significantly better attitudes at post-test than teachers in the F2F format, but no condition differences were found for mathematical achievement. Participants in the blended learning condition valued the flexibility of course structure and control over instructional pace but expressed that time management and decreased physical presence of the instructor were challenges. These findings suggest important elements to consider when designing blended learning courses.