Preprocessing of imagery time series is needed in order to carry out crop vegetative cycles analysis. Automatic normalization is a very interesting tool in the atmospheric correction process of satellite image time series in contrast to the radiative models. Thus, the purpose of this article is to ascertain the impact on the spectral-temporal profile of soybean crops using normalization through the multivariate alteration detection (MAD) technique during the 2004/2005 soybean harvesting season in Brazil. The normalized difference and greenness vegetation indices (NDVI/GVI) were selected to represent the temporal spectral profile. Five images were used for this study and all images were corrected for the atmospheric effect through the MAD technique, using the 5S radioactive transfer model. As the main outcome, it was noticed that normalization caused a negative impact on the spectral curves analysed, smoothing their shapes and distorting the crop growth curve.33516051626