The purpose of this study is to find models that best represent the software development process of IBM Federal Systems Division (FSD). The primary area of research is to evaluate existing effort estimating models and to develop a model best suited to IBM FSD's environment. The method of evaluation is an analysis of three existing models and their performance on a database of project information. The models used in this study included the Basic COCMO model by Boehm, the COPMO model by Thebaut and Putnam's macroestimating model. After extensive evaluation a hybrid effort estimating model was developed, and proven to be the best predictor of effort. / M.S.