Abstract: In the present code for building structure design, empirical load combination rules are commonly used, which generally results in non-objective design combination load values and does not provide an approach to evaluate the probability of an actual combination load to exceed a prescribed load value. Assuming that loads are stationary binomial processes, combinations of multi correlated or uncorrelated loads are studied. Considering the correlation of loads, the parameters of load combination random process are determined. The probabilistic distribution function of load amplitude of stationary binomial process for load combination is approached by the maximum entropy method, and the cumulative probabilistic distribution of maximum load combination is deduced. Further the load combination factors, which provide an approach to consider load combinations in actual design, are calibrated. Finally the load combination method presented in this paper is verified by a Monte Carlo simulation experiment. The loads by various empirical load combination rules are evaluated using the method developed in this paper.