Globally, neonatal mortality accounts for nearly half of under-five mortality, and intrapartum related events are a leading cause. Despite the rise in neonatal resuscitation (NR) training programs in low- and middle-income countries, their impact on the quality of NR skills amongst providers with limited formal medical education, particularly those working in rural primary health centers (PHCs), remains incompletely understood.This study evaluates the impact of PRONTO International simulation training on the quality of NR skills in simulated resuscitations and live deliveries in rural PHCs throughout Bihar, India. Further, it explores barriers to performance of key NR skills. PRONTO training was conducted within CARE India’s AMANAT intervention, a maternal and child health quality improvement project. Performance in simulations was evaluated using video-recorded assessment simulations at weeks 4 and 8 of training. Performance in live deliveries was evaluated in real time using a mobile-phone application. Barriers were explored through semi-structured interviews with simulation facilitators.In total, 1342 nurses participated in PRONTO training and 226 NR assessment simulations were matched by PHC and evaluated. From week 4 to 8 of training, proper neck extension, positive pressure ventilation (PPV) with chest rise, and assessment of heart rate increased by 14%, 19%, and 12% respectively (all p ≤ 0.01). No difference was noted in stimulation, suction, proper PPV rate, or time to completion of key steps. In 252 live deliveries, identification of non-vigorous neonates, use of suction, and use of PPV increased by 21%, 25%, and 23% respectively (all p < 0.01) between weeks 1-3 and 4-8. Eighteen interviews revealed individual, logistical, and cultural barriers to key NR skills.PRONTO simulation training had a positive impact on the quality of key skills in simulated and live resuscitations throughout Bihar. Nevertheless, there is need for ongoing improvement that will likely require both further clinical training and addressing barriers that go beyond the scope of such training. In settings where clinical outcome data is unreliable, data triangulation, the process of synthesizing multiple data sources to generate a better-informed evaluation, offers a powerful tool for guiding this process.