Abstract: To solve the problem of not good enough target detection results while applying present hyperspectral remote sensing band selection methods, a novel band selection method based on target detection-Two Step Band Selection(TSBS) was proposed. Primarily, hyperspectral bands were initially selected and sorted via typical band selection method. Then according to the target detection results of different band subsets after successively accumulating each band, the initially selected bands were reselected again. Finally, the band subset with better target detection result was obtained. The proposed method could not only inherit the advantages of typical methods, but also had higher application pertinence and simple implementation due to its direct target detection-based band selection. Experimental results show that TSBS method has better general applicability. Both hyperspectral data dimensionality reduction and improvement of target detection result can be realized effectively.