One of the most common methods of obtaining diagnostic information about a child's language production is to collect and analyze a spontaneous language sample. Usually, this sample is collected in a clinical setting by a speech-language pathologist. Because children's language production seems very sensitive to situational variables, there is some question about the representativeness of language samples obtained in clinical settings. This study compared language samples obtained in the home by the mothers with those obtained in the clinic by speech-language pathologists. Subjects were three- to five-year-old children who had been referred for speech and language diagnostics. Results were that the children's mean-length-of-utterance (MLU) scores were better for the home sample, although there was no difference between their Developmental Sentence Scores (DSS) for the home sample and the clinic sample. The home samples yielded higher MLU stage placements (Brown, 1973) and higher estimated language ages on the DSS (Lee, 1974) for most of the children. Clinical implications of the results are discussed.