Methods to infer cloud thermodynamic phase (ice or water) are investigated usingnmultispectral imagery. An infrared (IR) trispectral algorithm using the 8.52-, 11-, and 12µm bands [Ackerman et al., 1990; Strabala et al., 1994] forms the basis of this work and will be applied to data from the Earth Observing System (EOS) Moderate-Resolution Imaging Spectroradiometer (MODIS) instrument. Since the algorithm uses IR bands, it can be applied to either daytime or nighttime data and is not sensitive to the presence of cloud shadows. A case study analysis is performed with a MODIS airborne simulator (MAS) scene collected during the Subsonic Aircraft Contrail and Cloud Effects Special Study (SUCCESS) on April 21, 1996. The scene under scrutiny is quite complex, containing low-level broken water clouds, cirrus cells, cloud shadows, subvisual cirrus, and contrails. The IR trispectral algorithm is found to be less accurate in (1) regions where more than one cloud phase type occurs and (2) regions of thin cirrus overlying a lower-level cloud layer. To improve the phase retrieval accuracy in these areas of difficulty, additional bands located at 0.65, 1.63, and 1.90µm are incorporated. Radiative transfer (RT) calculations are performed to simulate the MAS bands using the cirrus and water cloud models detailed by Baum et al. [this issue]. The RT calculations are performed for singlelayer cirrus- and water-phase clouds as well as for the case of thin cirrus overlying a lower-level water droplet cloud. Both modeled results and application of the theory to a case study suggest that the cloud thermodynamic phase retrieval accuracy can be improved by inclusion of the visible and near infrared bands.