ABSTRACT Under favorable conditions, digitally recorded sonic log waveforms can be computer processed to produce a suite of logs including velocity and attenuation for both compressional and shear waves. Basic signal processing procedures can be applied to these waveforms in order to extend the range of conditions over which shear wave logs are obtainable. Low-pass frequency filtering effectively reduces compressional noise and "dip" filtering reduces energy reflected or converted at interfaces. Velocity filtering can also be used to discriminate against source-generated noise, provided the spatial sampling is adequate. Shear wave velocity analysis is accomplished by maximization of the cross-semblance between windows selected on the basis of predicted shear wave arrival times. This is done first on low-pass filtered waveforms to eliminate the possibility of cycle skipping and to reduce dispersion. The velocity log obtained in this way can be refined by iteration; i.e., the bandwidth is increased on the high end, the velocity aperture and correlation windows are narrowed, and the process is repeated.