The scattering signature in various microwave frequencies from atmospheric ice particles can be directly related to surface rainfall rates. From AMSU-B millimeter wavelength channels, we found the detection of light precipitation can be significantly improved due to an increasing sensitivity of scattering to smaller ice particle sizes. This study will demonstrate our operational global precipitation algorithm using AMSU millimeter wavelength channel measurements and possible applications of our methods for future global precipitation mission (GPM). In the retrieval algorithm, we first derive cloud ice water path (IWP) and ice particle effective diameters (De) simultaneously using AMSU-B measurements at 89 and 150 GHz. Then, the retrieved ice water amount is converted into the surface rainfall rate through an IWP and rainfall rate relationship developed from cloud model results. A screening procedure is also developed and applied to separate non-precipitating ice clouds from precipitating ones. In addition, the AMSU measurements at 183 GHz water vapor channels are tested for delineating the vertical distribution of frozen hydrometeors. According to the cloud ice distribution, precipitation systems are grouped into several categories so that separate relationships between IWP and rain rate can be used for deriving the surface rain rate. The retrieved rainfall rates are compared with radar and gauge data. Monthly mean precipitation products from AMSU are also compared with that from SSM/I. It is shown that the algorithm accurately depicts both localized rainfall events and rainfall patterns on the large scale.