Maintenance consumes the majority of software's lifecycle resources. Factors affecting the maintenance phase are of practical concern to systems developers and users, but only a few of those factors are under direct control of the project manager. Modeling formalisms are generally touted for their impacts on the development of software; however their impact on maintenance is less discussed and less clear. The question addressed in this research is what impact the choice of modeling formalism has on system maintenance activity. Modeling formalisms are used to capture, analyze and present systems. Major modeling formalisms include process, data, and object-oriented (OO). While process and data formalisms tend to focus more on system processes or data respectively, OO is intended to more evenly balance the treatment of both. Sixty organizations provided data on the use of modeling formalisms across the organization and data on 112 software development projects. The results show how the choice of a modeling formalism has significant differential implications for software maintenance. Despite other research results to the contrary, the results reported here lead us to the conclusion that the choice of an OO modeling formalism had the best impact on maintenance