Objective: This study sought to characterize change mechanisms that underlie gastrointestinal (GI) symptom improvement in IBS patients undergoing two dosages of CBT for IBS as compared to a nondirective education/support (EDU) condition. Method: Data were collected in the context of a large clinical trial that randomized 436 Rome III-diagnosed IBS patients (Mage = 41, 80 % female) to standard, clinic-based CBT (S-CBT), a largely home-based version with minimal therapist contact (MC-CBT) or Education/Support that controlled for nonspecific effects. Outcome was measured with the IBS-version of the Clinical Global Improvement scale that was administered at Week 5 and 2-week posttreatment (Week 12). Potential mediators (IBS Self-efficacy (IBS-SE), pain catastrophizing, fear of GI symptoms, and treatment alliance were assessed at Weeks 3, 5, and 8 during treatment with the exception of treatment expectancy that was measured at the end of Session 1. Results: IBS-SE, a positive treatment expectancy for symptom improvement, and patient-therapist agreement on tasks for achieving goals mediated effects of CBT early in treatment (rapid response, RR) and at posttreatment. Notwithstanding their different intensities, both CBT conditions had comparable RR rates (43%-45%) and significantly greater than the EDU RR rate of 22%. While pain catastrophizing, fear of GI symptoms, and patient-therapist emotional bonding related to posttreatment symptom improvement, none of these hypothesized mediators explained differences between CBT and EDU, thereby lacking the mechanistic specificity of IBS-SE, task agreement, and treatment expectancy. Conclusion: Findings suggest that CBT-induced GI symptom improvement may be mediated by a constellation of CBT-specific (IBS-SE) and nonspecific (task agreement, treatment expectancy) processes that reciprocally influence each other in complex ways to catalyze, improve, and sustain IBS symptom relief. (PsycInfo Database Record (c) 2021 APA, all rights reserved).