For preschool children with autism, individual behavioral interventions are among the best-tested treatments. However, they are rarely implemented in special education preschools. We observed one-to-one behavioral interventions formally and informally delivered by staff (N = 51) in 12 classrooms across three preschools for children with autism, aged 3-6 years, in a major US city. We estimated associations between the use of one-to-one intervention and classroom characteristics including staff-student ratio, professional role composition, and frequency of challenging child behaviors. A small number of classroom characteristics explained considerable portions of outcome variance: 23% for formally delivered one-to-one interventions and 41% for informally delivered interventions. The number of individually assigned personal care aides in the classroom was negatively correlated with less formal delivery of one-to-one intervention. Classroom challenging behavior was positively associated with formal delivery of one-to-one interventions. Interventionist's professional roles and the number of children in the class accounted for the largest amounts of variance in informal intervention delivery. Staff training, clarifying professional roles, setting performance expectations for personal care aides and other classroom team members, and reducing class size may represent promising implementation targets. Findings suggest caution around task-shifting policies that transfer clinical functions from more highly trained to less highly trained staff.