A major topic of debate is whether children with autism spectrum disorder should be educated in inclusive or specialized settings. We examined the feasibility and preliminary effectiveness of delivering the Group-Early Start Denver Model to children with autism spectrum disorder in inclusive versus specialized classrooms. We randomly assigned 44 preschoolers with autism spectrum disorder to receive the Group-Early Start Denver Model across one school calendar year in classrooms that included only children with autism spectrum disorder or mostly children who were typically developing. Blind-rated indicators of teaching quality showed similar results across settings, which were above the local benchmark. Children showed improvements across blinded proximal measures of spontaneous vocalization, social interaction, and imitation and across distal measures of verbal cognition, adaptive behavior, and autism symptoms irrespective of intervention setting. Mothers of participants experienced a reduction in stress irrespective of child intervention setting. Across both settings, age at intervention start was negatively associated with gains in verbal cognition. Delivery of Group-Early Start Denver Model in an inclusive setting appeared to be feasible, with no significant differences in teaching quality and child improvements when the program was implemented in inclusive versus specialized classrooms.