The goal of this study is to develop a fatality probability function associated with injury severity and age, which could provide a useful method to estimate the fatality rate accurately. Seven types of logistic regression models were taken into consideration. The results of the estimations from the 7 logistic regression models were compared to select the best fit regression model by using the data from US accident statistics (NASS‐CDS:​ National Automotive Sampling System Crashworthiness Data System) in year 2001. In addition, the constant coefficients of the model best fit to the year 2001 data were replaced by regression functions as a function of year to develop a model incorporating the effect of the year change. The following results were found:​ 1) the best fit regression model was a function of maximum AIS (Abbreviated Injury Scale) of each body region and age;​ 2) the accuracy of the regression model was improved by applying regression functions of the coefficients as a function of the year;​ and 3) the regression functions of the coefficients show that the fatality rate decreases with each progressing year.