This paper presents the results of a hybrid estimation model for perfective software maintenance applied in a real context. The resulting model integrates heuristics and parametric methods to improve the maintenance estimation. One of the main objectives of this model is the inclusion of functionalities that provide zero function points. These functionalities can be supported by the model through impact points. Likewise, the micro function points and weighted risks have been included to provide a more realistic estimate. Finally, the model was implemented in a SME where the estimates were very approximates to the real effort. As result, the model could be implemented in a small organization with a stable maintenance process and effort deviation was 11% after applying the model.