This paper proposes a framework for assessing quantitatively the error-proneness of computer program modules. The model uses an information theory approach to derive an Error Proneness Index that can be used in a practical way. Debugging and testing take at least 40% of a software project's effort, but do not uncover all defects. While current research looks at identifying problem-modules in a program, no attempt is made for a quantitative error-proneness evaluation. By quantitatively assessing a module's susceptibility to error, we are able to identify error prone paths in a program and enhance testing efficiency. The goal is to identify error prone paths in a program using Genetic Algorithms. This increases software reliability, aids in testing design, and reduces software development cost.