Objective The aim of this study was twofold: (a) Investigate whether therapists are consistent in their use of therapeutic techniques throughout supportive–expressive therapy (SET) and (b) Examine the bi‐directional relation between therapists' use of therapeutic techniques and the working alliance over the course of SET. Method Thirty‐seven depressed patients were assigned to 16 weeks of SET as part of a larger randomized clinical trial (Barber, Barrett, Gallop, Rynn, & Rickels, ). Working Alliance Inventory‐Short Form (WAI‐SF) was collected at Weeks 2, 4, and 8. Use of therapeutic interventions was rated by independent observers using the Multitheoretical List of Therapeutic Interventions (MULTI). Intraclass correlation coefficients assessed therapists' consistency in use of techniques. A cross‐lagged path analysis estimated the working alliance inventory‐ Multitheoretical List of Therapeutic Interventions bidirectional relation across time. Results Therapists were moderately consistent in their use of prescribed techniques (psychodynamic, process‐experiential, and person‐centred). However, they were inconsistent, or more flexible, in their use of “common factors” techniques (e.g., empathy, active listening, hope, and encouragements). A positive bidirectional relation was found between use of common factors techniques and the working alliance, such that initial high levels of common factors (but not prescribed) techniques predicted higher alliance later on and vice versa. Conclusion Therapists tend to modulate their use of common factors techniques across treatment. Additionally, when a strong working alliance is developed early in treatment, therapists tend to use more common factors later on. Moreover, high use of common factors techniques is predictive of later improvement in the alliance.