Community-based newborn care (CBNC) has been implemented in Ethiopia across the maternal, neonatal, and child health continuum of care with the goal of lowering newborn mortality. However, neonatal mortality rate in Ethiopian is among the highest in the world. Why neonatal mortality remains high in the face of such effective interventions is the issue. As a result, the authors claim that it is unknown whether the planned intervention is carried out effectively or not. The purpose of this study was to investigate the fidelity of community-based newborn care intervention and its implementation drivers. Multicenter community-based mixed method study was employed on 898 postpartum women, 16 health extension workers (HEWs) and 10 health posts to evaluate CBNC intervention fidelity. Structured questionnaire and facility audit checklist was used to collect quantitative data. In-depth interview technique was used to explore lived experiences of HEWs on CBNC implementation. CBNC intervention fidelity was computed as a composite index of the product of program coverage, frequency and contents. Multilevel linear regression model with adjusted β-coefficients at P-value of 0.05 and a 95% confidence interval (CI) were used to declare a significant relation between CBNC intervention fidelity and its implementation drivers. Interpretative phenomenological analysis was employed for qualitative data analysis. CBNC intervention fidelity was found to be 4.5% (95% CI: 3.6-5.4) with only two women received the intervention with full fidelity. The overall CBNC intervention coverage was 38.4% (95% CI: 35.2-41.6). Only 8.1% and 1.5% of women received all CBNC interventions with recommended frequency and content, respectively. HEWs knowledge of danger sign was significant facilitator while lack of: health center's feedback, related short-term training, health development army support, health center staff's technical assistance to HEWs and shortage of medical equipment supply were barriers for CBNC intervention fidelity. In conclusion the CBNC intervention fidelity was too low in this study. This indicates that CBNC intervention package was not implemented as envisioned implying an implementation gap. All implementation drivers were poorly implemented to result in improved fidelity and intervention outcomes.