Learning support in mathematics (LSM) is widely available at colleges and universities through center-based peer-tutoring and classroom-based programs such as Peer-Assisted Learning (PAL). The purpose of this study was to experimentally introduce a six-week PAL-program in selected first-year undergraduate math courses, in order to determine its effect on students' math anxiety, performance, attitudes, and behaviors toward college math, with a particular focus on highly math-anxious (HMA) students. An explanatory mixed-methods sequential design (Creswell, 2014) was employed in order to explore the research questions first quantitatively and then qualitatively. Student survey results, course grades, and tutoring records were used to address the research goals of Phase 1. In Phase 2, interviews with highly math-anxious students were analyzed using a grounded theory approach (Charmaz, 2006; Strauss & Corbin, 1998), providing context and meaning for the quantitative results of Phase 1. Quantitative analysis showed no significant effects of the PAL-treatment on math anxiety, performance, attitudes, or help-seeking behavior. However, students' average perceived value of mathematics improved significantly across the entire study sample. Phase 2 interviews provided evidence that the improvement in affect was due primarily to excellent teaching and not the PAL program. Interviews also revealed a ���Disconnect� between students' knowledge of what it takes to succeed in math and their own behaviors. Recommendations for future research include exploring implementation strategies to maximize the effect of a new PAL program, examining the qualities of excellent teaching to see how they might influence PAL programs, and engaging stakeholders (students, teachers, and PAL-leaders) to create a culture where all students recognize the value of LSM and do not hesitate to ask for help.