At present, the etiologic yield in community-derived samples of young children with an autistic spectrum disorder is not known. To address this question, all young children (under 5 years of age) referred for an initial assessment to ambulatory pediatric neurology or developmental pediatric clinics at a tertiary university center over an 18-month period for a suspected developmental delay were prospectively identified. Specific diagnostic testing was left to the discretion of the evaluating physician. In all, 50 children with an autistic spectrum disorder were assessed. Detailed history or physical examination was informative with respect to suggesting the possibility of an underlying etiology in a minority (10/50,20%). Genetic studies (FMR-1, karyotype), electroencephalography (EEG), and neuroimaging were carried out in a majority (42/50, 34/50, and 33/50, respectively) of the children, for the most part on a screening rather than an indicated basis (31/42, 34/34, and 28/33, respectively). Etiologic yield was low (1/50,2%), with only a single child identified with a possible Landau-Kleffner variant on sleep EEG tracing. The results suggest an evaluation paradigm with reference to etiologic determination for young children with autistic spectrum disorder that does not presently justify metabolic or neuroimaging on a screening basis. Recurrence risk and treatment implications, however, suggest that strong consideration be given to genetic (FMR-1, karyotype) testing and EEG study despite a relatively low yield. (J Child Neurol 2001;16:509-512).