In this paper we suggest a special generalized proportional hazard model for counting process and discuss its statistical problems for a large sample study. The counting process under consideration of this paper has three jumping times. Under the assumption for the model of this paper, we will show that the data observed in the time interval after the second jumping time provide no information for estimating the true parameter θ(subscript (0)). Secondly, we suggest an estimating function of θ(subscript (0)), and use it to provide the information calculations. Finally, we discuss the asymptotic consistency, asymptotic normality and asymptotic efficiency respectively for the maximum partial likelihood estimator of θ(subscript (0)).