To determine the etiologic yield in young children with developmental delay referred to sub-specialty clinics for evaluation. Over an 18-month period, all children less than 5 years of age referred to the ambulatory pediatric neurology or developmental pediatrics clinics of the Montreal Children's Hospital for initial evaluation of a suspected developmental delay were enrolled. Features evident on history or physical examination were determined at intake as were the laboratory tests (and their rationale) requested by the evaluating physicians. Six months post initial assessment, detailed chart review was undertaken to determine if an etiology was found and the basis for such a determination. Bivariate and multivariate logistic regression was used to test for associations between factors present at intake and successful ascertainment of an underlying etiology. Two hundred and twenty-four children met study criteria. Etiologic yield varied across childhood developmental delay subtypes, and was 44/80 for global developmental delay [GDD] (55%), 13/22 for motor delay [MD] (59.1%), 3/72 for developmental language disorders [DLD] (4.2%), and 1/50 for autistic spectrum disorders [ASD] (2%). For GDD, the presence of historical features or findings on physical examination was associated with greater likelihood for successful etiologic determination with the following items significant in multiple logistic regression analysis; microcephaly, antenatal toxin exposure, focal findings. For MD, physical findings or the co-existence of a cerebral palsy symptom complex predicted a successful search for etiology. For both groups, the severity of the delay did not predict etiologic yield. For both groups, a small number of etiologic categories accounted for the majority of diagnoses made. Etiologic yield in childhood developmental delay is largely dependent on the specific developmental delay subtype. Paradigms for systematic evaluation of this common child health problem can be suggested, however they await validation.