The present study attempted to test the reliability of a new method of recording verbal behavior in a free-play preschool setting. Six children, three normal and three speech impaired, served as subjects. Videotaped records of verbal behavior were scored by two experimentally naive observers. The results suggest that the system provides a means of obtaining reliable records of both normal and impaired speech, even when the subjects exhibit nonverbal behaviors (such as hyperactivity) that interfere with direct observation techniques.