A system of orthogonal functions, known as Walsh functions, which assume only the values +1 and −1, are presented as a tool appropriate for the analysis of well-log data. Many of the characteristics of these rectangular waveforms, such as discrete transitions in signal level, make them ideal for processing borehole data. Using the generalized Fourier transform, the basic equations for the Walsh transform and Walsh power spectrum are developed showing how the energy in a given signal is distributed among these rectangular wave components. As a first application of the Walsh transform to logging data, a method of computeraided rock boundary identification is developed and demonstrated on a set of well logs from a continental basalt sequence. This technique provides a fast, simple yet accurate means of dividing well-log data into groups of measurements corresponding to different rock units.