Software cost estimation is vital for the effective control and management of the whole software development process. Currently, Constructive Cost Model (COCOMO II) is the most popular tool for estimating software cost. It uses Lines of Code and Function Points to assess software size. However, these are actually implementation details and difficult to estimate during the early stage of software development. The Entity Relationship (ER) model is well used in conceptual modeling (requirements analysis) for data-intensive systems. In this article, we explore the use of ER model for the estimation of software cost. A new term, Path Complexity, is proposed. Based on Path Complexity and other factors, we built a multiple regression model for software cost estimation. The approach has been validated statistically through system data from the real industry projects.