Wavelet signal decomposition techniques applied to fathometer echoes were used for seafloor classification. Fathometer echoes were deconvolved with their source signals to yield transfer functions representative of the seafloor. These transfer functions were then expanded onto damped complex exponential wavelet bases. In this case a discrete implementation of the continuous wavelet transform was used. This technique allowed the signals be decomposed into distinct modes by localizing signal component energies in time and frequency. Modal features were then analyzed for clues to the physical makeup of the seafloor. Techniques were developed for extracting modal features from the wavelet signal expansions. These modal features include modal densities, center frequencies, bandwidths, duration, and most importantly, decay rates. The modal features were analyzed for clues to the physical makeup of the sea bottom. Available information was exploited to invert for internal sediment structure, particularly sediment density and the presence of trapped bubbles. Feature-based bottom maps were generated allowing classification of the seafloor. [Work supported by NRL/SSC under the MTEDS program.]