Play therapy has been employed by therapists to alleviate children's emotional stress. Most research dealing with this type of intervention has been in the form of case studies, focusing on therapy outcome rather than on the therapeutic process. The process of one child in individual play therapy was analyzed through the use of time-series analysis. A 5-year-old boy, referred because of his social isolation and withdrawal in a classroom setting, participated in 8 weekly 45-minute play therapy sessions. Stimulus-response verbalizations were established as the dependent variable. A total of 21 30-minute measurements were taken twice weekly over an 11-week period: 6 pre-intervention time points and 15 post-intervention time points. A stochastic model was fit to the data; the time-series was stationary with no significant serial correlation present in the first five lags of the post-intervention data. Although a linear regression yielded a significant treatment effect, examination of the residuals using Q statistic revealed serial dependence, negating the model viability. An alternative statistical model was formulated which produced a significant treatment effect and confirmed the goodness-of-fit of the model. The results suggest that play therapy was a viable intervention for this child. (Author/NRB)