The Streamlined Antiretroviral Therapy Initiation Strategy (START-ART) study found that a theory-based intervention using opinion leaders to inform and coach health care providers about the risks of treatment delay, provision of point of care (POC) CD4 testing machines (PIMA) and reputational incentives, led to rapid rise in ART initiation. We used qualitative research methods to explore mechanisms of provider behaviour change.We conducted in-depth interviews (IDIs) with 24 health care providers and nine study staff to understand perceptions, attitudes and the context of changes in ART initiation practices. Analyses were informed by the Theoretical Domains Framework.Rapid dissemination of new practices was enabled in the environmental context of an existing relationship based on communication, implementation and accountability between Makerere University Joint AIDS Program (MJAP), a Ugandan University-affiliated organization that provided technical oversight for HIV service delivery at the health facilities where the intervention was implemented, and a network of health facilities operated by the Uganda Ministry of Health. Coaching carried out by field coordinators from MJAP strengthened influence and informal accountability for carrying out the intervention. Frontline health workers held a pre-existing strong sense of professional identity. They were proud of attainment of new knowledge and skills and gratified by providing what they perceived to be higher quality care. Peer counsellors, who were not explicitly targeted in the intervention design, effectively substituted some functions of health care providers; as role models for successful ART uptake, they played a crucial role in creating demand for rapid ART initiation through interactions with patients. Point of care (POC) CD4 testing enabled immediate action and relieved providers from frustrations of lost or delayed laboratory results, and led to higher patient satisfaction (due to reduced costs because of ability to initiate ART right away, requiring fewer return trips to clinic).Qualitative data revealed that a multicomponent intervention to change provider behaviour succeeded in the context of strong institutional and individual relationships between a University-affiliated organization, government facilities, and peer health workers (who acted as a crucial link between stakeholders) and the community. Fostering stable institutional relationships between institutional actors (non-governmental organization (NGOs) and ministry-operated facilities) as well as between facilities and the community (through peer health workers) can enhance uptake of innovations targeting the HIV cascade in similar clinical settings.