In this paper, we propose a compression algorithm for hyperspectral images based on the wavelet transform using adjacent information. A characteristic of the SPIHT algorithm is that it provides information on the locations of significant coefficients of transformed images. On the other hand, there exist high correlation between spectral band images of hyperspectral data. Thus, by using the location information of significant pixels, it is possible to efficiently compress adjacent band images by allocating all available bits to encode the magnitudes of significant pixels. However, some significant pixels would be inevitably missed, resulting in noticeable errors. In this paper, we propose to encode these missed significant pixels to improve coding efficiency. Experiments suggest that taking care of missed significant pixels provides improved performance.