All the known dip-moveout (DMO) algorithms that are not integral methods require the seismic data to be sorted in regularly sampled constant-offset sections. In contrast, the dip-moveout method proposed here can be applied directly to recorded shot profiles and thus can handle data that cannot be sorted in regular constant-offset sections.The definition of the shot-DMO operator is analogous to that of the dip-moveout operator for constant-offset sections. The two operators have impulse responses with the same projection on the zero-offset plane, i.e., the stacking plane; therefore, the application of dip moveout in constant-offset sections or in shot profiles gives the same stacked section. Dip moveout transforms shot profiles to zero-offset data, to which any poststack migration can be applied.The shot-DMO operator is space-variant and time-variant; thus direct application of the operator would be computationally expensive. Fortunately, after a logarithmic transformation of both the time and the space coordinates, the operator becomes time-invariant and space-invariant; then dip moveout can be performed efficiently as a multiplication in the Fourier domain. Shot dip moveout is also a useful tool for improving the accuracy of residual velocity analysis performed after the DMO process.Field-data examples show that the shot-profile dip-moveout method yields stacked sections similar to those from Hale's (1984) dip moveout for constant-offset sections.