Human mortality, which reflect the deaths’ extent, is one of the key research of Population Science and Population Economics. Accurately mortality forecasting can lay the foundation of pricing longevity risk bonds. Based on Lee-Carter model, this paper considers mortality correlations and investigates the long-run equilibrium of mortality rates between China mainland and Taiwan province for mortality forecasts. Differing from the traditional ARIMA model which is based on the limited data of China, the paper proposes a VECM model for the mortality time index forecasts after the co-integration test. Minimum mean square prediction errors (MSPE) is used for criteria, our results show that the forecasting based on the VECM model is better. Mortality rates under the multi-region framework can provide important reference for further study of pricing a multi-region longevity bonds.