The most challenges problems in hyperspectral images processing are the huge amount of data volume and the high correlation between bands. Bands selection technique is one of the common approaches to overcome these issues in order to deal with many applications. However, there are two main issues arising from bands selection which must be addressed as the amount of required bands and the choice of the optimal criterion needed to be used in selecting bands. To deal with these two issues, this demonstration presents a progressive bands selection, which performs progressive band dimensionality and reduction through band prioritization scores calculated by combining various statistical criteria. We have applied the proposed approach on real hyperspectral data and the obtained results show the effectiveness compared to other criteria used in the progressive bands selection.