Regional velocity functions based on sonobuoy solutions were used to predict four layer thicknesses at three Leg 29 drill sites with an average error of 7%. A computer program (based on an improved R (sub S) -R (sub 1) technique) to determine velocity gradients from individual sonobuoy records was applied successfully to data from the New Zealand Plateau and central Tasmanian Sea. Direct comparisons with sonobuoy solutions at or very near drill sites indicated no significant difference between observed and computed data. Predicted velocity functions and observed mean velocities (based on well-identified reflectors) were compared with Leg 29 sonic probe data. It was observed that the coring process had very little effect on the sonic velocity of sediments taken from below an overburden of about 300 m thickness, but that sonic velocities measured in sediments with less overburden were much too low. The largest discrepancies occurred in material taken from zones with 200 m of overburden, where the measured values averaged 180 m/sec less than the predicted values.