Abstract
To estimate the corrective software maintenance effort, we must know the factors that have the strongest influence on corrective maintenance activities. In this study, we have analyzed activities and effort of correcting 810 software defects in one Norwegian software company and 577 software defects in another. We compared the defect profiles according to the defect correction effort. We also analyzed defect descriptions and recorded discussions between developers in the course of correcting defects in order to understand what led to the high cost of correcting some types of defects. The study shows that size and complexity of the software to be maintained, maintainers?' experience, and tool and process support are the most influential cost drivers of corrective maintenance in one company, while domain knowledge is one of the main cost drivers of corrective maintenance in the other company. This illustrates that models for estimating software corrective maintenance effort have to be customized based on the defect profiles and cost drivers of each company and project to be useful.