Summary.The postneonatal mortality rate in three southern regions of New Zealand was 8.1 per 1000 live births for 1979–1984. The possibly preventable postneonatal mortality rate was 7.1 per 1000 live births and data on 377 possibly preventable deaths and 936 randomly selected controls were used to develop a two‐stage risk‐scoring system. Logistic regression analysis was undertaken separately on two sub‐samples of the total sample. The data were used to evaluate three other previously published scoring systems. The four variables used in our birth score (mother's age, parity, marital status, and birth weight) have all appeared in other scoring systems, and this score could identify a group of infants in southern New Zealand with an estimated 1.7% mortality rate in the first year of life. It is suggested that currently the only valid use of risk‐scoring in this region is for the definition of a high‐risk population for the purpose of evaluating potential intervention