One hundred thirty-seven students in 12th grade physics classes participated in a quasi-experimental study comparing the jigsaw classroom method of cooperative instruction with traditional direct instruction. While no differences were found between the two conditions for physics achievement gains, the results revealed differences in students' experience of the three basic needs (autonomy, competence, and social relatedness as posited by self-determination theory of learning), in self-reported cognitive activation, and in degree of intrinsic motivation. Path analyses showed that the basic needs partially mediated the effects of method of instruction on cognitive activation and intrinsic motivation. Increases in feelings of competence with cooperative learning were associated with better performance in physics. When controlling for competence, however, direct instruction had a facilitating effect on physics performance. Four aspects of students' personal learning characteristics (previous knowledge, academic self-concept in physics, academic goal orientation, uncertainty orientation) were assessed. Method of instruction was found to interact with self-concept: students with low academic self-concept profited more from cooperative instruction than from direct instruction because they experienced a feeling of greater competence.