Software cost estimation is one of the most challenging works due to the fact that it requires detailed information about each stage of the project, including the project scope, project requirements, software development methodology and the required resources. Often, this information is not known until the final stage of project while the cost of project needs to be estimated at the initial stage. There are many different ways to estimate software costs and each way has its own characteristics and purpose according to each project profile. The purpose of this research is to propose a framework for developing a software cost estimation model for software modification project based on a relational matrix of a project profile and software cost using an analogy estimation method and multiple regression analysis to estimate the amount of software modification list and the amount of effort need for the new project from the data collection of the previous project.