In my thesis I focus on providing a foundation of data on whichdecision makers can base refactoring decisions. For this, I examine therelationship between software complexity and maintenance eort. Tomake the data a stronger basis for refactoring decisions, I present anew approach of correlating le metrics to maintenance eort, whereI look at the relation between changes in le metrics over multiplereleases and changes in the maintenance eort spent on these les. Ido this using a broadened and, more complete notion of maintenanceeort. I measure maintenance eort in 4 ways: the amount of lines ofcode that had to be changed to resolve tasks, the amount of discus-sion that tasks generated, the amount of atomic changes to a le thatwere required to resolve a task, and the amount of bugs per month.To test this framework, I extracted data from 3 open source projects,where I measured the variation of both complexity and maintenanceeort, using this new notion of eort, over multiple releases, and in-vestigated their correlation. I found that 21 of the tested metrics weresignicantly correlated to the eort measures, where complexity basedmetrics and incoming propagation cost show the highest correlation.Of the proposed measures for maintenance eort, the amount of dis-cussion to resolve an issue shows the highest correlation to the chosenmetrics.