Prestack computations for cross‐hole data are relatively expensive, as they are for prestack surface survey data. It is therefore of interest to develop methodologies for modeling and processing stacked cross‐hole data. In this context, stacking is over sources, not midpoints. Modeling with a line source produces data that are equivalent (by Huygen’s principle) to those obtained by stacking over a line of point sources. Reverse‐time finite‐difference migration may be applied to the resulting stacked section by generalizing the excitation‐ time imaging condition for a point source to a line source. Illustrations include successful applications to both synthetic data and scale‐model data.