Context. The use of integral field units (IFUs) is now commonplace at all major observatories offering efficient means of obtaining spectral as well as imaging information at the same time. The IFU instrument designs are complex and spectral images have typically highly condensed formats, therefore present challenges for the IFU data reduction pipelines. In the case of the VLT VIMOS-IFU, a fringe-like pattern affecting the spectra well into the optical and blue wavelength regime as well as artificial intensity variations, require additional reduction steps beyond standard pipeline processing. Aims. We propose an empirical method for the removal of the fringe-like pattern in the spectral domain and the intensity variations in the imaging domain. We also demonstrate the potential consequences of a failure to correct for these effects for any subsequent data analysis. Here we use the example of deriving stellar velocity, velocity dispersion, and absorption line-strength maps for early-type galaxies. Methods. For each spectrum that we reduce with the ESO standard VIMOS pipeline, we derive a correction spectrum by using the median of the eight surrounding spectra as a proxy for the unaffected, underlying spectrum. This method relies on the fact that our science targets (nearby early-type galaxies) cover the complete field-of-view of the VIMOS-IFU, have spectral properties that vary gradually with position and that the exact shape of the fringe-like pattern is nearly independent and highly variable between neighboring spatial positions. Quadrant-to-quadrant intensity variations are corrected for in an independent step. Results. We find that the proposed correction methods for the removal of the fringe-like pattern and the intensity variations in VIMOS-IFU data-cubes are suitable to permit a meaningful data analysis for our sample of nearby early-type galaxies. Since the method relies on the scientific target properties it is unsuitable for any general implementation in the pipeline software for VIMOS.