Rates of participation in intervention research have not been extensively studied within autism spectrum disorder. Such research is important given the benefit of early intervention on long-term prognosis for children with autism spectrum disorder. The goals of this study were to examine how family demographic factors predicted treatment attendance and adherence in a caregiver-mediated randomized controlled trial targeting core deficits of autism spectrum disorder, and whether treatment attendance and adherence predicted outcome. In all, 147 caregiver-child dyads from a low-resourced population were randomized to in-home caregiver-mediated module or group-based caregiver education module treatment. Treatment attendance, adherence, and outcome (time spent in joint engagement) were the primary outcome variables. The majority of families who entered treatment (N = 87) maintained good attendance. Attendance was significantly predicted by socioeconomic status, site, and treatment condition. Families in caregiver-mediated module reported lower levels of treatment adherence, which was significantly predicted by site, condition, caregiver stress, and child nonverbal intelligence quotient. Dyads in caregiver-mediated module had significantly longer interactions of joint engagement, which was significantly predicted by an interaction between treatment attendance and condition. Overall, the results from this study stress the importance of considering demographic variables in research design when considering barriers to treatment attendance and adherence.