Although hyperspectral images provide abundant information about objects, their high dimensionality also substantially increases computational burden. Dimensionality reduction offers one approach to Hyperspectral Image (HSI) analysis. Currently, there are two methods to reduce the dimension, band selection and feature extraction. In this paper, we present a band selection method based on Independent Component Analysis (ICA). This method, instead of transforming the original hyperspectral images, evaluates the weight matrix to observe how each band contributes to the ICA unmixing procedure. It compares the average absolute weight coefficients of individual spectral bands and selects bands that contain more information. As a significant benefit, the ICA-based band selection retains most physical features of the spectral profiles given only the observations of hyperspectral images. We compare this method with ICA transformation and Principal Component Analysis (PCA) transformation on classification accuracy. The experimental results show that ICA-based band selection is more effective in dimensionality reduction for HSI analysis.