Coherent radar imaging (CRI) is used in an attempt to overcome the angular resolution limitation of conventional single-station radars and is used to image the horizontal structure inside the resolution volume. This recently developed technique has been successfully applied to radar observations of the ionosphere as well as the lower atmosphere. However, no statistical analysis of the robustness of the various techniques has been presented to date. In this work, three CRI techniques are reviewed: Fourier-based, Capon's, and maximum entropy (MaxEnt) methods. The Fourier-based method is the simplest of the three algorithms but has inherent resolution limitations. Although quite different in nature and performance, both Capon's and MaxEnt methods can be posed as constrained optimization problems. A statistical comparison of performance of the three CRI techniques, using various receiver configurations and two distinct cases of scattering structure, is made using simulated data. The results show that the MaxEnt method exhibits the best performance in the case of aspect-sensitive scattering with a broad characteristic. In the localized scattering case, however, Capon's method shows superior performance for signals with high signal-to-noise ratio (SNR), but MaxEnt method outperforms all methods for low SNR. In general, both Capon's and MaxEnt methods are able to reproduce the gross characteristics of the scattering media under observation. ¿ 2000 American Geophysical Union