In this paper an algorithm for retrieving precipitation from Special Sensor Microwave/Imager (SSM/I) satellite microwave data is presented for rainfall occurring over both ocean and land. This algorithm is basically derived from the results of a microwave radiative transfer model of plane-parallel rain clouds and combines the emission and scattering regimes. The features of the algorithm are as follows: (1) a linear function of horizontally polarized brightness temperature at both 19.35 and 85.5 GHz is used as a parameter to relate rainfall rate, which shows no saturation for rainfall rates up to 50 mm/h; (2) the relationship between the linear function and the rainfall rate is closer to linearity than the relationship between brightness temperature of a single channel and rainfall rate, and therefore the beam-filling problem is less serious than for single-channel algorithms; (3) the algorithm includes a correction to a beam-filling problem that was made on the basis of observational evidence; (4) the threshold for precipitation is derived from the SSM/I data themselves, which includes climatological information; and (5) the algorithm is less sensitive to freezing level height and surface temperature than emission-based algorithms. Validation of the algorithm is made using surface radar and rain gauge data. The validation shows that rainfall retrievals over ocean are in a satisfactory agreement with the surface observations. Over land, heavy rainfall rate can be detected although the disagreement is significant. ¿ American Geophysical Union 1992