More and more organizations invest resources to improve cost prediction and management during software development. Several cost estimation models with the aspects of parameters and requirements have been proposed to forecast software cost. However, the intended objectives of those models lie not only in forecasting but also in management. Without controllable factors those models are simply grouped as forecasting models rather than prediction models. If a model does not have at least one controllable factor, it will not directly provide effective action to affect the predicted outcomes. The process performance model (PPM) is a statistical collection of the relationships among attributes of processes and specific outcomes that can be used to construct cost prediction models. In this paper, we describe how to use practical leverage and actual data to establish an empirical process performance model with statistical regression techniques to predict software cost and manage cost using controllable factors. With regression analysis, the proposed model can provide outcome prediction to manage projects cost more confidently.