A standard approach to the selection of children with specific language deficit was devised. The approach was based on a current definition of specific language deficit that depends heavily on exclusion criteria. Children with known language deficit who had hearing impairment, cognitive deficit, neurologic deficit, or emotional or behavioral disorder were excluded. Both the exclusion criteria and the criteria dealing with required extent of language impairment were specified in considerable detail. A total of 132 language-impaired children aged 4-8 1/2 years were assessed. Criteria for selection of a matched group of normal-language children were also specified. The language-impaired children were referred to the project by speech and language clinicians in accordance with a set of broadly defined criteria. Contrary to expectation, less than one third of the children from this group met the more precisely defined criteria employed in the project. In addition, those who were selected did not form a homogeneous group but showed considerable variation in expressive versus receptive language and in articulation skill. The selection approach and its effectiveness for the purposes of clinical research and intervention studies are discussed.