To improve the outcome of task fMRI analyses, we compared the result maps obtained following a modulated analysis pipeline with despiking and CompCor added as denoising steps, with those obtained following the standard analysis pipeline. The data was taken from 5 studies. We focused on the overlap between the obtained individual findings, between the individual and group results and between the findings of 2 fMRI sessions. Our results revealed a reduction in false results in the ventricles, but failed to show an increased repeatability of fMRI results after the denoising. In conclusion, denoising prior to the statistical analysis seems advisable.