Identifying root cause of code smells is crucial in enabling proper and efficient software maintenance. Desired are computationally less intensive and easy to implement solutions. Embodiments herein provide a method and system for code analysis to identify causes of code smells. The method provides considerable automation during software maintenance by analyzing detected code smells of two subsequent versions of an application by categorizing the code smells, filtering significant code smells from the detected code smells, mapping the significant code smells between two versions, generating a customized PDG for the mapped functions and performing the isomorphic subgraph mapping for computing a function difference. The computed function difference provides added part of the code or removed part or region of the code of the application, which is cause of the significant added code smells, which can then be provided for manual analysis to identify the root causes for efficient software maintenance.