The statistical distribution of random variables is absolutely necessary in many engineering practices, such as reliability analysis, fatigue life test. When building a statistical model, in general, a hypothesis was proposed based on statistical data of samples and experience of the engineer, who took the responsibility. Therefore the hypothesis of the statistical model would be influenced by the opinion of the engineer. Based on maximum entropy concept, statistical model of complicated random variables was presented, and a parameter estimation method of the distribution function was also proposed. Maximum entropy statistical models were used to describe service loads of engine and clutch of a light truck, and rotating speed. The result showed that this model is feasible and valid.