The potential of ground-based polarimetric cloud radar measurements to improve information about cloud microphysics is evaluated. Using the discrete dipole approximation (DDA), backscattering at 95 GHz (3.16 mm) has been computed for solid hexagonal columns and plates, hollow cylinders, and two types of stellars. A randomly oriented compact column (a=1) represents a first step toward considering polycrystals at millimeter wavelengths. Crystal sizes up to 2 mm in maximum dimension and random orientation with maximum particle dimension in one plane have been considered, while the incidence angle is varied from 0¿ to 90¿. Thus the effect of shape, size, and orientation on radar observables as copolar reflectivity (Zhh), linear depolarization ratio (LDR), and differential reflectivity (ZDR) as well as difference reflectivity (ZDP) can be investigated. Technical radar aspects such as the potential of scanning and cross polarization isolation requirements are assessed as well as the benefit of alternative wavelengths considered for cloud radar application (35, 140, 220 GHz). At 220 and 140 GHz, distinct resonance features are found, complicating the interpretation of radar measurements at those frequencies. At 95 GHz the effects of shape or orientational variation on Zhh are of the same order and much smaller than the effect of crystal size. On the other hand, LDR is governed by the crystal shape and orientation but almost independent of size. The results indicate that if crystals have a preferred orientation, LDR is a useful observable for differentiating between the two major crystal types of columnar and planar shapes. However, it is shown that the corresponding cross-polar backscatter intensities are theoretically and technically difficult to achieve. ZDR and ZDP involve only copolar intensities. If the lowest applicable antenna elevation angle is about 45¿ for a ground-based measurement, these parameters are promising for discriminating horizontally aligned pristine crystals from randomly oriented ones or irregular aggregates. ¿ 1999 American Geophysical Union