Two-dimensional migration of seismic time sections by a digital computer has been performed by a variety of techniques, ranging from computer-simulated hand migration to a computer simulation of elastic wave propagation from the surface receiver positions back into the subsurface. Past processes have utilized the normal moveout equation, whereas the subject process is an application of its spatial derivative. 'Dip-domain' migration pertains to special sections where all events have the same space-time slope, and the migration at each slope is approximated by a trivial function of the subsurface rms velocity and the associated vertical two-way traveltime. The actual process is to (1) dip decompose a seismic section into so-called 'dip blocks' through the applications of appropriate dip-discriminating coherency functions, (2) migrate the individual dip blocks, and (3) recompose the migrated dip components (dip blocks).The dip-discriminating coherency functions are computed from a combination of an 'algebraic' and an 'absolute' dipping spatial mix of the seismic time section. The input seismic data are not themselves spatially mixed on the dip blocks, but, rather, they are reduced in amplitude by the coherency functions for the dip blocks of which they are not a coherent component. Migration formulas are developed in the Appendix for both spatially invariant and variable subsurface velocity functions.The subject process has some computational advantages. In addition, improved signal-to-noise ratios have consistently been obtained due to the dip bandpass (fan) filtering inherent to the process. Synthetic and field seismic examples are included.