Distribution transformer condition assessment and remaining life prediction is an important prerequisite to carry out condition-based maintenance. This paper, based on the transformer health index (HI) of each year, puts forward a kind of distribution transformer life combination forecast model based on the theory of cross entropy. In residual life model, failure probability density function and cross entropy theory are introduced to dynamically determine the weight of the improved gray theory and Markov prediction algorithm. Considering the change of the shape parameter and scale parameter of failure probability density function in the life cycle of the distribution transformer, it realizes the evaluation of the full life cycle of transformer’s health index and determines the distribution function of the health index. Actual data verify the effectiveness and reasonability of the model. It is more in line with the actual situation compared to a single prediction method.