Although previous studies have shown favorable results with early intensive behavioral treatment (EIBT) for children with autism, it remains important to replicate these findings, particularly in community settings. The authors conducted a 3-year prospective outcome study that compared 2 groups: (1) 21 children who received 35 to 40 hours per week of EIBT from a community agency that replicated Lovaas' model of EIBT and (2) 21 age- and IQ-matched children in special education classes at local public schools. A quasi-experimental design was used, with assignment to groups based on parental preference. Assessments were conducted by independent examiners for IQ (Bayley Scales of Infant Development or Wechsler Preschool and Primary Scales of Intelligence), language (Reynell Developmental Language Scales), nonverbal skill (Merrill-Palmer Scale of Mental Tests), and adaptive behavior (Vineland Adaptive Behavior Scales). Analyses of covariance, with baseline scores as covariates and Year 1-3 assessments as repeated measures, revealed that, with treatment, the EIBT group obtained significantly higher IQ (F = 5.21, p = .03) and adaptive behavior scores (F = 7.84, p = .01) than did the comparison group. No difference between groups was found in either language comprehension (F = 3.82, p = .06) or nonverbal skill. Six of the 21 EIBT children were fully included into regular education without assistance at Year 3, and 11 others were included with support; in contrast, only 1 comparison child was placed primarily in regular education. Although the study was limited by the nonrandom assignment to groups, it does provide evidence that EIBT can be successfully implemented in a community setting.