Objective: Psychotherapies like Acceptance and Commitment Therapy (ACT) are thought to target multiple clinical outcomes by intervening on multiple mechanistic process variables. However, the standard mediation approach does not readily model the potentially complex associations among multiple processes and outcomes. The current study is one of the first to apply network intervention analysis to examine the putative change processes of a psychotherapy. Methods: Using data from a randomized trial of ACT versus minimally-enhanced usual care for anxious cancer survivors, we computed pre- to post-intervention (n = 113) change scores on anxiety-related outcomes (general anxiety symptoms, cancer-related trauma symptoms, and fear of cancer recurrence) and putative processes of the intervention (experiential avoidance, self-compassion, and emotional approach coping). We estimated a network model with intervention condition and change scores as nodes.Results: Contrary to the expectation that intervention effects would pass indirectly to outcomes via processes, network analysis indicated that two anxiety-related outcomes of the trial may have acted as primary mechanisms of the intervention on other outcome and process variables.Conclusions: Network intervention analysis facilitated flexible evaluation of ACT’s change processes, and offers a new way to test whether change occurs as theorized in psychotherapies.