Mentoring is frequently discussed as a viable approach for improving the educational and social outcomes of students at risk for learning or behavioral problems. However, little data-based evidence beyond case study has been presented regarding the effectiveness of mentoring for students at risk. Here, 16 students with emotional/behavioral disorders (EBD) in two classrooms at an alternative day school were paired with undergraduate preservice teachers who served as mentors. Using PAND (Percentage of All Non-Overlapping Data), data indicate behavior change for 13 of the 16 students with improvement for 6 and a range of effects sizes. This data indicate that mentoring relationships with certain vulnerable populations (such as students at risk for or with EBD) may have mixed effects. (Contains 3 tables.)