**Background:** Realist evaluation examines what works for whom, under what circumstances, and how. Despite numerous public health interventions in low- and middle-income countries (LMICs), few studies have explicitly detailed how program implementation triggers health system strengthening. The "Abiye" (Safe Motherhood) program was initiated in 2009 in Ondo State, Nigeria, to address the four delays associated with maternal mortality: delay in seeking care, delay in reaching care, delay in receiving care, and delay in referral. The program had demand components (free services, community health extension workers called Health Rangers, strengthened referrals) and supply components (facility renovation, equipment, human resources). This study aimed to unveil the "black-box" of program implementation and health system strengthening by identifying contextual factors, mechanisms, and developing Context-Mechanism-Outcome (CMO) configurations.
**Methods:** This was qualitative research structured along realist domains (Context, Mechanism, Outcome). An Initial Program Theory (IPT) was developed from document reviews, key informant interviews with 12 policymakers/program developers/implementers, and peer-reviewed literature. The IPT was validated through qualitative data collection: 10 key informant interviews (with the immediate past governor, commissioner of health, five directors, and local government coordinators), 28 in-depth interviews (with six Health Rangers, six facility health workers, 15 women who delivered before and during the program, and one traditional birth attendant), and six focus group discussions (with mothers who delivered in 2015–2016 and older women aged 60+). Three local government areas (one per senatorial district) and six wards were selected. Thematic analysis using Braun and Clarke's six-phase guideline was applied, with deductive mapping to realist domains. Seven themes and 19 subthemes emerged.
**Key Results:** Thirteen contextual factors were identified under five principal areas: situational (poor background maternal health, political/policy entrepreneur), structural (geographical, health workforce, health facilities), design (financial, community, data management, sustainability), cultural (cultural belief, traditional birth attendant, Agbebiye program), and international community. Most factors played enabling roles; cultural beliefs and traditional birth attendants played inhibiting roles; financial context played both enabling and inhibiting roles (enabling when funding was regular, inhibiting when salaries became irregular). Eight mechanisms were elicited: political mechanisms (high-level policy entrepreneurship, commitment, legal nudge via CEMDOS law, but no enabling law for sustainability), funding mechanisms (central funding system, financial protection, but demotivation from irregular payments), human resource mechanisms (motivation via multiple incentives, knowledge acquisition, task shifting, central pooling), Health Ranger mechanisms (bonding, perceived trust), facility/service delivery mechanisms (motivation, service efficacy), data management mechanisms (accountability, tracking), community mechanisms (mobilization, good health-seeking behavior, but poor health-seeking behavior in some riverine communities), and TBA/Agbebiye mechanisms (carrot-and-stick approach). Health system strengthening was a key outcome: the state health budget increased from 2.9% (2009) to 11% (2014); consultants increased from 4 (2007) to 72 (2016); eight obstetricians and six pediatricians were employed; all drugs and materials were provided free; BEOC and CEOC services were available 24/7; the HMIS was repositioned; and the CEMDOS law was enacted. Unintended outcomes included skewed prioritization/funding toward Abiye at the expense of other programs, and an increased state health wage bill. Six CMO configurations were developed, leading to a new middle-range theory: "Provision of sustainable free quality maternal care to all pregnant women by motivated health professionals within a comprehensive health design in fully engaged communities, with a background of strong political will and prioritization, would improve utilization of the health facilities and skilled delivery thereby sustainably reducing maternal mortality in Ondo State."
**Clinical Implications:** The study provides a framework for maternal mortality reduction programs in LMICs with six key elements: (a) policy championing and political prioritization, (b) comprehensive health system design, (c) strong data monitoring, (d) strengthened community systems and ownership, (e) elimination of non-skilled deliveries, and (f) adequate funding. The "Agbebiye" program (vocational training for traditional birth attendants) and the tracking system are novel mechanisms that could be adapted in other settings. The findings emphasize that legal sustainability mechanisms must be incorporated at the planning stage to prevent program decline when political leadership changes. The study fills a gap in realist evaluation literature from Africa and provides evidence that can be adapted for policymaking in public health interventions in LMICs.