The efficiency and reliability analysis is an important issue for the design and operations management of transfer lines. In this paper, a method combining discrete event simulation (DES) and metamodel is presented to predict the performance of transfer lines with unreliable workstations and finite buffers. The DES can conduct computer experiments for transfer lines with different system parameters under the Matlab environment. The metamodel is constructed by using stochastic Kriging integrated with Latin hypercube sampling (LHS), which can predict the responses from a limited set of simulation results. The main advantages of the proposed method are its flexibility and ease of use. Also, it is the first time to investigate stochastic Kriging opportunities in modeling the performance of transfer lines. Finally, numerical cases indicate that the proposed procedure can achieve higher prediction accuracy than analytical method.