Reverse time migration (RTM) received its name because its inventors were literally running the time axis of their modeling programs backward and computationally sending waves back into a model of the Earth to make images. By extrapolating waves in time, rather than along the vertical direction, waves could go all directions rather than just within some angle of the vertical. Initial RTM images were astounding for the steep dips they rendered that other methods missed. Equally importantly, because the wave propagation in an RTM algorithm is “complete”, the velocity model didn't have to be smoothed; complex models were no problem as they were with Kirchhoff methods, the main competing steep-dip imaging method. These advantages are easily understood in the context of poststack migration and the exploding reflector model. Prestack RTM is slightly more complicated, requiring simulations of both source and receiver wavefields, but the advantages of extrapolating the time axis remain. Even though this “time reversal of the physics” is conceptually appealing, practical application of RTM languished over the years since its inception in the early 1980s. We think this was partly due to cost, partly due to issues such as internal scattering leading to image artifacts and the difficulties and cost of producing high-quality image gathers, and, finally, partly due to the insufficiency of the data we acquired. We think it's fair to say that RTM is only truly justified on 3D data; so, RTM had to wait for the 3D revolution.