Self-explanation is an important constructive cognitive process that helps students learn in such a way that they can flexibly transfer their knowledge to solve novel problems (Chi, Bassok, Lewis, Reimann, & Glaser, 1989). However, research has not addressed what leads students to spontaneously self-explain, in the absence of prompting. The present study experimentally manipulates student motivation (in terms of achievement goals) and measures what influence this has on self-explanation and transfer. Participants (N = 140) received goal framings that reflected either a mastery-approach goal (striving to develop one’s understanding), a performance-approach goal (an aim to outperform others), a performance-avoidance goal (avoid doing worse than others) or a no-goal control. This framing was applied to a set of learning and test tasks on basic statistics, which participants completed while thinking aloud. Results showed a benefit for a performance-avoidance condition in terms of both higher levels of self-explanation and transfer. This unexpected result is discussed in terms of theories of motivation and learning, and their potential impact on educational practice.