A method is presented for reducing travel time from multiple offset borehole seismic experiments to velocity-depth structure. This technique, which treats simultaneously data from any number of depths in the borehole in addition to surface data, is based on the linear inversion scheme of Dorman and Jacobson (1981). Given the parameters &zgr;(p)=T(p)+pX(p) and &tgr;(p)=T(p)-pX(p), the method solves for depth to specified slownesses, assuming linear gradients (in velocity) between slowness values. A practical limiation on the use of surface travel time measurements to resolve velocity-depth structure is the necessity of assuming a surface velocity. This is not necessary for the case of borehole data. For a borehold receiver the velocity at the depth of the receiver can be obtained from the slope of the inflection point of the travel time curve. Thus a direct measure of the uppermost velocity of a section can be obtained from the travel time data of a shallow borehole receiver. Estimation of velocities from the inflection points of deeper receivers improves the resolution of the velocity-depth function, which would be obtained from surface data alone. The technique is applied to data from three boreholes seismic experiments in oceanic crust. The experiments were carried out in the western Atlantic (crustal age ~110 m.y.), the Gulf of California (crustal age ~1 m.y.) and the Costa Rica Rift (crustal age ~6 m.y.). All three experiments show relatively high upper crustal velocities (>4.0 km/s), suggesting that layer 2A is not present even in the very young crust. All sites had over 100 m of sediment thickness, and it is postulated that sediment thickness and sediment premeability, not merely age, govern the velocity of the upper crust by accelerating the cementation of fractures and cracks.