We present initial results of applying statistical control techniques to the massive maintenance processes of a software organization rated at the CMM level 3. In particular, we present results from an empirical study conducted on a large massive adaptive maintenance project. In a previous study we analyzed the correlation between the maintenance size and productivity metrics and produced models to estimate the costs of a project conducted according to the adopted maintenance processes [13]. In this paper we analyze data about the single phases of the process and in particular the distribution of the effort among the phases and the causes of variations.