The purpose of this study was to investigate the effects of proximity fading and task breaks on responding during small group Direct Instruction in mathematics with students with ASD and the extent to which students demonstrate academic accuracy and generalization of responding. Using a multiple probe across participants design no functional relation was identified between group responding and proximity fading and task breaks. All students demonstrated improvement in mathematics performance from pre-test to posttest. Social validity indicated "strongly agree" for group responding instruction and "strongly disagree" for mathematics curriculum with regards to pacing of instruction and amount of language demands placed on participants. Future research on optimal group size, prerequisite skills required for effective group participation, and longer duration of DI mathematics instruction on skill acquisition is suggested.