In a previous study, we compared the effects of just over one year of intensive behavior analytic intervention (IBT) provided to 29 young children diagnosed with autism with two eclectic (i.e., mixed-method) interventions (Howard, Sparkman, Cohen, Green,& Stanislaw, 2005). One eclectic intervention (autism programming; AP) was designed specifically for children with autism and was intensive in that it was delivered for an average of 25-30 h per week (n = 16). The other eclectic intervention (generic programming; GP) was delivered to 16 children with a variety of diagnoses and needs for an average of 15-17 h per week. This paper reports outcomes for children in all three groups after two additional years of intervention. With few exceptions, the benefits of IBT documented in our first study were sustained throughout Years 2 and 3. At their final assessment, children who received IBT were more than twice as likely to score in the normal range on measures of cognitive, language, and adaptive functioning than were children who received either form of eclectic intervention. Significantly more children in the IBT group than in the other two groups had IQ, language, and adaptive behavior test scores that increased by at least one standard deviation from intake to final assessment. Although the largest improvements for children in the IBT group generally occurred during Year 1, many children in that group whose scores were below the normal range after the first year of intervention attained scores in the normal range of functioning with one or two years of additional intervention. In contrast, children in the two eclectic treatment groups were unlikely to attain scores in the normal range after the first year of intervention, and many of those who had scores in the normal range in the first year fell out of the normal range in subsequent years. There were no consistent differences in outcomes at Years 2 and 3 between the two groups who received eclectic interventions. These results provide further evidence that intensive behavior analytic intervention delivered at an early age is more likely to produce substantial improvements in young children with autism than common eclectic interventions, even when the latter are intensive.