The Peer Assisted Learning (PAL) program at Sacramento State was established in 2012 with one section supporting introductory chemistry. The program now serves 17 courses with high rates of students who receive a D or an F or withdraw (DFW) from the course in biology, chemistry, mathematics, physics, and statistics; the program enrolls approximately 1,400 students annually. Adapting the Peer-Led Team Learning model, PAL facilitators do not teach, tutor, or even confirm answers; they ask scaffolding questions, provide encouragement, and ensure that all group members participate in problem-solving. Each PAL section is an optional credit-bearing course that supplements the targeted parent science, technology, engineering, and mathematics (STEM) course. In this article, we assess the efficacy of the program in terms of student success in the parent course. As PAL is an opt-in program, we employ propensity score matching techniques to account for confounding factors. Our analysis shows that the mean course grade point average is 1.98 for matched nonparticipants and 2.40 for matched PAL participants, indicating that the program provides an average bump of 0.42 points in the parent course. We consider data from more than 25,000 students, and our propensity score analysis uses more than 10,000 students (4,519 PAL and 5,814 non-PAL) for whom appropriate matches could be found.