Background: There has been slow progress in the reduction of neonatal mortality, with almost all deaths occurring in low-income countries. The global target to reduce neonatal mortality to 12 deaths per 1000 live births by 2030 requires tremendous effort, because most countries in sub-Saharan Africa are unlikely to reach this target by 2030 (WHO, 2020a). Neonatal mortality in Nigeria is one of the highest in the world. Education packages targeted at health workers and focused on providing knowledge and skills around the essential components of newborn care have been shown to reduce neonatal death. This study designed and evaluated the impact of an educational intervention on neonatal care and survival at the Jos University teaching hospital, Jos, Nigeria. Methods: A multiphase mixed methods design was adopted for the study and was underpinned by Kirkpatrick’s framework for evaluating the effectiveness of a training program. Kirkpatrick proposed four levels of evaluating training: Reaction (Level one), learning (Level two), behaviour (Level three) and results (Level four). Phase One determine the current practices and training needs of health workers. Data in Phase One were collected using interviews and questionnaire surveys completed by 21 health workers. In Phase Two, the WHO package was modified using the findings of Phase One. In Phase Two, the modified package was implemented and Kirkpatrick’s Levels one and two (reaction and learning) were evaluated. Both qualitative and quantitative data were collected using questionnaires before and after the study and through group interviews after training. Thirty-eight (38) health workers participated in this phase. Phase Three assessed behaviour change as a result of the training. Non-participatory observation was used to assess behaviour. Data were collected using a tool developed by Sobel et al. (2010) to observe 30 deliveries in the labour ward two months after training. In Phase Four, it was determined whether there had been any improvements in neonatal survival. Data regarding neonatal deaths six months before and six months after the intervention were collected from the hospital records department. Qualitative data were analysed in NVivo to conduct thematic analysis while quantitative data were analysed using SPSS. Results: The study findings indicated that the quality of newborn care in the current setting needs urgent improvements. There was evidence of early separation of neonates from their mothers after birth, interrupted skin-to-skin contact and delayed breastfeeding initiation. Furthermore, routine oral and nasal suctioning, early cord clamping, and unnecessary separation had been reported pre-training. Participants were generally satisfied with the training implemented to improve newborn carepractices, and the mean post-training score (21.42±1.08) demonstrating an improvement from the pre-training score (17.95±3.08), which represents 13.35% increase in general performance. The correlation of performance before and after training was positive and statistically significant (0.427; p=0.008). This indicated a similarity in the ranking order of pre-training and post-training scores. Findings also suggested a reasonable improvement in newborn care when compared with pretraining practices. Skin-to-skin contact was maintained for 25 babies (80.6%) and was sustained until breastfeeding was initiated for 20 deliveries (64.5%). The umbilical cord was clamped an average of 47 seconds after delivery. Hands were washed before care was performed on the newborn, with new gloves being used for all babies. Cord care was performed with alcohol for approximately 77% of babies, while 23% were cleaned with chlorhexidine. Babies were immediately dried and dressed, and no baby was placed on a cold surface. The body temperature of 19 babies (61.3%) was checked within the first two hours after delivery. There was a decline in neonatal mortality rate from 53/1000 live births in the six months preintervention to 37/1000 live births six months post-intervention. Independent T test analysis of the day of neonatal death demonstrates a significant difference in mean (P=0.00, 95% CI - 5.629; -7.447- -4.779). The mean values for day of neonatal death in the pre and post intervention periods were 3.35 ±2.2 and 9.46 ±3.2 days after birth respectively. This implies that most of the deaths in the pre intervention period occurred in the early neonatal period while most deaths that occurred in the post intervention period occurred in the late neonatal period. Conclusion: This is the first study to apply Kirkpatrick’s framework in a paediatric setting in Nigeria, and to involve participants in planning the intervention. Moreover, the study has contributed to reducing early neonatal mortality, harmful and inappropriate practices during newborn care, and it has contributed to the professional development of health workers by providing an effective model of training implementation and evaluation. It has also potentially reduced early newborn death. The study findings indicate that the quality of newborn care in the current setting needs urgent improvements. This was evident prior to the intervention, where early separation of neonates from their mothers after birth interrupted skin-to-skin contact and breastfeeding initiation. Furthermore, routine oral and nasal suctioning, early cord clamping, and unnecessary separation had been reported pre-training. The modified package improved the capacity of health workers to care for neonates and resulted in significant improvements in the quality of newborn care. Furthermore, there was a slight reduction in early neonatal mortality in the post-intervention period. It is recommended that health workers in the current setting require support from the hospital management in the supply of essential consumables and developing policy that will promote evidence-based neonatal care.