The mortality prediction of the elderly above retirement age is the basis of pension actuarial valuation and longevity risk measurement．In order to solve the problem of small amount and large fluctuation of mortality data of people above retirement age in mainland China，this paper proposes a Logistic multi-population model for the elderly mortality model based on a single population Cairns-Blake-Dowd ( CBD) model．By adding more relevant population data information to predict the mortality rate of the elderly population in China，this paper selects the gender specific mortality data in Taiwan and carries out joint modeling with the gender specific mortality rate in mainland China．It is found that compared with single population CBD model，the Logistic multi-population mortality model has a better fitting and long-term coherent prediction effect．
