PurposeThe criteria for autism spectrum disorder (ASD) were revised in the fifth edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM). The objective of this study was to compare the sensitivity and specificity of DSM-IV-Text Revision (DSM-IV-TR) and DSM-5 definitions of ASD in a community-based sample of preschool children.MethodsChildren between 2 and 5 years of age were enrolled in the Study to Explore Early Development-Phase 2 (SEED2) and received a comprehensive developmental evaluation. The clinician(s) who evaluated the child completed two diagnostic checklists that indicated the presence and severity of DSM-IV-TR and DSM-5 criteria. Definitions for DSM-5 ASD, DSM-IV-TR autistic disorder, and DSM-IV-TR Pervasive Developmental Disorder Not Otherwise Specified (PDD-NOS) were created from the diagnostic checklists.Results773 children met SEED2 criteria for ASD and 288 met criteria for another developmental disorder (DD). Agreement between DSM-5 and DSM-IV-TR definitions of ASD were good for autistic disorder (0.78) and moderate for PDD-NOS (0.57 and 0.59). Children who met DSM-IV-TR autistic disorder but not DSM-5 ASD (n = 71) were more likely to have mild ASD symptoms, or symptoms accounted for by another disorder. Children who met PDD-NOS but not DSM-5 ASD (n = 66), or vice versa (n = 120) were less likely to have intellectual disability and more likely to be female. Sensitivity and specificity were best balanced with DSM-5 ASD criteria (0.95 and 0.78, respectively).ConclusionsThe DSM-5 definition of ASD maximizes diagnostic sensitivity and specificity in the SEED2 sample. These findings support the DSM-5 conceptualization of ASD in preschool children.