Atmospheric radar imaging techniques have shown promise in revealing the fine-scale structure of the atmosphere within the resolution volume of the radar. Enhanced resolution can be obtained in both angle and range by using spaced receivers and shifted frequencies, respectively. The distinct techniques have been termed coherent radar imaging (CRI) for angular resolution enhancement and range imaging (RIM) for radial resolution improvement. Because of the mathematical similarities between CRI and RIM it is possible to derive a generalization of both techniques. In this work, the three-dimensional (3-D) imaging technique, which uses multiple receivers and multiple frequencies simultaneously, is developed for the first time. Three-dimensional imaging has the advantage of mitigating the limitations of beam width as well as pulse width of a conventional radar to simultaneously improve both angular and range resolution. It is shown that CRI and RIM are special cases of 3-D imaging. The mathematical problem is formulated as an inverse problem with solutions provided by the Fourier, Capon, and maximum entropy (MaxEnt) methods. These three 3-D imaging methods are verified and statistically tested through numerical simulations. ¿ 2001 American Geophysical Union