Background Neonatal mortality has remained a major public health concern in low- and middle-income countries. Early warning scoring (EWS) systems minimize hospital mortality by providing rapid and efficient care; however, their effectiveness in low- and middle-income countries is unknown. Method This study is a retrospective analysis of neonates admitted to the University College Hospital in Nigeria between January and December 2019. We evaluate the accuracy of significant modified early warning scores (sMEWS) in predicting neonatal in-hospital mortality and associated risk factors. Results The majority (254; 68.5%) of the 371 admitted newborns were late preterm to term, and 54.7% were males; 173 (48.7%) required resuscitation at admission in the emergency department (ED). One hundred and thirty-seven had sMEWS, which decreased to 22 (5.9%) 72 h post-admission. Univariable analysis shows that sMEWS at admission were associated with sepsis, perinatal asphyxia, and the necessity for resuscitation. Multivariable logistic regression indicated that outborns (adjusted odds ratio [AOR]: 2.218; 95% confidence interval [CI], 1.125–4.371), need for resuscitation at admission (AOR: 5.501; CI, 3.350–9.034) and persistence of sMEWS 72 h post-admission (AOR: 9.693; CI, 2.181–43.079), were associated with sMEWS. Significantly associated with excess mortality were sMEWS at admission (AOR: 3.530; CI, 1.721–7.240) and sMEWS 72 h post-admission (AOR: 5.931; CI, 5.944 –64.385). Conclusions: The sMEWS performed moderately well as an EWS in a neonatal emergency unit. Excess mortality was associated with the need for resuscitation in the ED and the persistence of sMEWS 72 h post-admission.