Certain introductory engineering courses have been termed gateway courses, due to high failure rates resulting in many students leaving engineering altogether. In one such course, thermodynamics, instructors intentionally formed peer-led study groups (PLSGs) with an undergraduate teaching aide (UGTA) playing a supportive facilitator role instead of providing direct instruction. PLSG students met one hour per week for eight weeks, over and above the standard teaching assistant-led recitation. The intervention resulted in statistically significant improvements in students' course grades, pass rates, and graduation rates compared to students who participated in the standard recitation only. PLSG earned statistically almost one full letter grade higher than students in the no-treatment group; they were also statistically more likely to have passed the course and to have graduated with their degree approximately one year after taking it. PLSG students were also compared to students in a control group who participated in an additional hour of the standard teaching assistant-led recitation each week. Students in the PLSGs performed at similar levels to students in the control despite receiving far less instruction and guidance from the teaching aide. Notably, no differences were found in the effectiveness of the intervention for transfer students, women, and racial/ethnic minoritized students, indicating that the intervention provides at least the same educational advantages to underrepresented groups as to non-underrepresented groups. These results support the implementation of peer-led student groups in thermodynamics and other upper-level gateway engineering courses. Preparation and training videos are available for those interested in implementing this intervention in their own courses: https://lth.engineering.asu.edu/reference-guide/peer-led-study-groups/.