Background: Autism is a devastating neurologic disorder for which little meaningful treatment is available. Although a highly unconventional notion, several strands of evidence suggest the possibility that in a subset of children with autism, certain pathologic gut flora may contribute to their symptomatology (perhaps via neurotoxin elaboration). We conducted a pilot prospective, open-label study to test this hypothesis. Methods: Twelve children (11 males, 1 female, mean age = 4.9y, range 3.5-6.75y) with delayed onset autism were studied. All other neurologic, genetic and metabolic diagnoses were excluded. Entry criteria included antibiotic exposure and subsequent chronic persistent diarrhea preceding the onset of their autistic features. Intervention was with oral vancomycin for 8 weeks. Several measures were used to assess and document change, including the Childhood Autism Rating Scale (CARS), the Developmental Profile II (DEV-II), and a subjective analog rating scale. These scales were completed by a teacher blinded to the intervention, a parent, and a psychologist at baseline and following the intervention. Results: Significant improvement was noted by all three observers on the CARS (2×3 ANOVA, F (1,4) = 8.01,p <.05). Striking changes in hyperactivity, attention, and interpersonal relatedness were consistently evident in analog ratings. Several children rapidly achieved new developmental milestones (e.g., toilet training) as noted on the DEV-II. Conclusion: The apparent clinical improvement after antibiotic intervention presents evidence to support the possible relationship between a gut flora alteration and autistic symptoms in a subset of children. Future research should include a randomized, double-blind, placebo-controlled intervention.