In this talk, we propose a hierarchical credibility approach to modeling multi-population mortality rates using data from the Human Mortality Database. Hierarchical credibility is used in a premium calculation with grouped data from different levels in a tree structure. Traditional mortality models for a single population estimate the model parameters with mortality data grouped into two levels (age and year). We group the mortality data for both genders of several developed countries into four levels (country, gender, age and year) or three levels (population, age and year), apply the hierarchical credibility to each level to better reflect the correlation of mortality rates among populations or genders/countries, and compare the forecasting performances between the model and some multi-population mortality ones.