This article describes how the concept of multiresolutionis used with cluster analysis of spectral data. Multiresolution analysis progressively increases the resolution ofa spectrum profile by adding levels of details containedin scales obtained from a discrete wavelet transform. Ateach resolution level a cluster analysis is performed onall the spectra profiles in the data set. This allows therelating of changes in the cluster pattern to various broadand narrow features in the spectral data profiles. Theanalysis also provides an approximate location of theimportant features in the original wavenumber domain.