Many software reliability growth models are proposed to be used in practice. However, most software reliability growth models suffer in the realistic software testing environment due to the unrealistic assumptions, such as perfect debugging, constant fault detection rate and regular changes. In fact, considering more reasonable assumptions in the reliability modeling may further improve the fitting and predictive power of software reliability growth models. It is affected by many factors, such as tester’s skill, test plans, testing tools and runtime environment. Thus, software debugging is an imperfect process. And software testing for getting fault data set is done under the assumption that user’s operation environment is the same as the testing one. However, in practice, it is exactly the same. This paper deals with a software reliability growth model which considers imperfect debugging and disagreement between operation environments. The better performance of proposed model is illustrated with fault data sets from software development project.