Objective To develop a pain assessment scale for Chinese neonates and to evaluate its validity and utility. Methods One hundred and twenty three neonates receiving arterial puncture for blood sample collection were recruited in a Neonatal Intensive Care Unit (NICU). Twenty of them were videotaped during arterial puncture, and 1 053 screenshots were taken every three seconds for developing the Neonatal Faces Pain Scale (NFPS). The other 103 neonates were also videotaped before, during, and after the arterial puncture procedure, for evaluating reliability, validity and consistency of the NFPS. The Neonatal Facial Coding System (NFCS) and the Neonatal Infant Pain Scale (NIPS) were used as the criteria. Fortythree NICU nurses were invited to evaluate the feasibility and clinical utility of the three scales. Results The NFPS included facial expression and physiological indicators, with a total score of 0 to 10 points. The interrater correlation coefficient and testretest reliability of the NFPS were 0.994 and 0.989, respectively. The Cronbach′s α coefficients of facial expression and physiological indicators were 1.000 and 0.717, respectively. The convergent validity coefficients ranged from 0.530 to 0.951 and the criterion validity coefficients were 0.952 and 0.944, respectively. The 95% limits of agreement in difference between the NFPS and NFCS were (-0.78, 1.47) and between the NFPS and NIPS were (-2.05, 1.52). Nurses scored the highest on the NFPS for feasibility and clinical utility than on the NFCS and NIPS (P<0.05), and 79.1% of them gave priority to use the NFPS. Conclusion The NFPS had good reliability, validity, and utility to assess procedural pain in neonates.