In the reliability assessment of products,this paper tried to make full use of a small amount of field data under natural storage, which might break through the restriction of using single pseudo-sample data in accelerated tests. The comprehensive reliability evaluation model,fusing exponential and Weibull natural storage field data and accelerated storage sample data,was built based on the integrating of relative entropy and the least square method. Simulation results indicate that the method can fuse two sources of information effectively,and when the amount of field data is small,the assessment of model is robust.