In this paper the coherence of retrieved cirrus microphysical and bulk properties using data from a satellite-based dual-viewing and multispectral instrument is tested using different ice crystal models. Radiance data from the dual-viewing Along Track Scanning Radiometer (ATSR-2) instrument is used to show that coherent retrievals are possible between nonabsorbing (visible) and infrared wavelengths if an appropriate ice crystal model is employed. The dominating crystal habit is estimated by finding the ice crystal model that best fits the dual-view 0.87 μm reflectance data. The ice crystal models tested are hexagonal plates, hexagonal columns, bullet-rosettes (six branched), and randomized polycrystals, all of which are assumed to be randomly oriented in space. Given the best fit crystal shape other cirrus properties, such as optical depth, crystal maximum dimension, and an estimate of ice water path, are retrieved by contrasting reflectance data at the wavelengths of 0.55 and 1.6 μm. To demonstrate probable retrieval errors in terms of optical depth and crystal maximum dimension, if the wrong crystal habit is applied, a tropical convective case in the western Pacific Ocean is used as a typical example. It is found that the more complex particles as represented by the bullet-rosette and randomized polycrystal best fit the ATSR-2 radiance data, while the pristine geometries represented by the hexagonal plate and column do not. These results indicate that phase functions that are relatively flat at backscattering angles should be employed in satellite remote sensing of cirrus. Moreover, if hexagonal plates or columns were assumed as the habit in the radiative transfer model, then this would lead to retrieval errors of about a factor of 2 for optical depth (overestimate) and crystal maximum dimension (underestimate). To test if the retrieved optical depth at the wavelength of 0.55 μm is coherent, the extinction optical depth at the wavelength of 10.8 μm is also retrieved to test for a one-to-one relationship between the two wavelengths. To validate this procedure the same test is applied to a near coincident aircraft and ATSR-2 midlatitude case study. ¿ 1999 American Geophysical Union