There is a need for a measure of children's playroom behavior to assess areas of functioning relevant to diagnosis, therapy process, and outcome, as discussed in clinical literature. This study concerned the development of a Play Therapy Observational Instrument designed to meet criteria specified as necessary for such a measure. Two nonclinicians were trained in the observation and scoring of 31 child behaviors. The observations were made on video tapes and transcripts of ongoing play therapy sessions. Results indicated that 13 of the behaviors were scored with a high level of interjudge reliability. These reliable scores formed three statistically valid and theoretically meaningful subscales: emotional discomfort, use of fantasy play as a coping method, and the quality of the child's interaction with the therapist. Children's scale scores were sufficiently stable within sessions and across two therapy sessions to indicate that they reflected individual differences. Relationships between scales were assessed and discussed in the light of theoretical controversies. Implications for future research were discussed.