This study presents normative data on the Boston Naming Test (BNT) for 382 normally developing children as a framework for evaluating what the BNT measures: word-retrieval or vocabulary. A Revised Children's BNT is proposed based on item analyses of the 212 boys' responses. Some BNT items were part of the boys' vocabulary at all ages, some entered their working vocabulary with age, some were affected by age and experience, and others were not part of the boys' working vocabulary at any age. Error analyses revealed that the pattern of errors in the boys differed from that of adult aphasics. Two error types, semantic and circumlocution, accounted for 82% of the boys' errors whereas three types characterize adult aphasics: semantic, phonemic, and circumlocution. The boys made semantic errors most frequently on age-related items; circumlocution errors were made on items that were not part of their working vocabulary. Suggestions for evaluating aphasic-like errors are presented.