Objective: Individuals’ use of self-statements reflecting self-criticism, a sense of inadequacy, and low self-worth has been linked to the incidence, severity, and recurrence of major depressive disorder [de Graaf, L. E., Huibers, M. J. H., Cuijpers, P., & Arntz, A. (2010). Minor and major depression in the general population: Does dysfunctional thinking play a role? Comprehensive Psychiatry, 51(3), 266–274. doi:10.1016/j.comppsych.2009.08.006; Riso, L. P., du Toit, P. L., Blandino, J. A., Penna, S., Dacey, S., Duin, J. S., … Ulmer, C. S. (2003). Cognitive aspects of chronic depression. Journal of Abnormal Psychology, 112(1), 72–80. doi:10.1037/0021-843X.112.1.72]. The present study used an observer-rated measure, the Structural Analysis of Social Behavior [Benjamin, L. S. (1974). Structural Analysis of Social Behavior. Psychological Review, 81(5), 392–425. doi:10.1037/h0037024], to examine patients’ self-directed communication over the course of psychotherapy.Method: Self-talk in early and late therapy sessions was examined using cases (N = 44) from the cognitive therapy arm of Jacobson and colleagues’ component study of cognitive therapy for depression [Jacobson, N. S., Dobson, K. S., Truax, P. A., Addis, M. E., Koerner, K., Gollan, J. K., … Prince, S. E. (1996). A component analysis of cognitive–behavioral treatment for depression. Journal of Consulting and Clinical Psychology, 64(2), 295–304. doi:10.1037/0022-006x.64.2.295, Jacobson, N. S., Dobson, K. S., Truax, P. A., Addis, M. E., Koerner, K., Gollan, J. K., … Prince, S. E. (2000). A component analysis of cognitive–behavioral treatment for depression. Prevention & Treatment, 3(1). doi:10.1037/1522-3736.3.1.323a]. We identified changes at treatment termination, and used hierarchical multiple regression to examine whether improvements in patient self-talk influenced treatment outcome up to 24 months post-termination.Results: Trends indicate that patients used friendlier and less critical self-statements at the end of treatment. Decreased self-critical behaviour was associated with fewer symptoms at the end of treatment and up to one year later; increased self-acceptance was linked to symptom improvement a year and a half after termination.Conclusion: Consistent with cognitive theory, reduced self-criticism was associated with better treatment outcomes. Longer-term improvement was linked to the development of friendlier and more accepting self-referential behaviour.