A technique is analyzed to mitigate the hostile effects of a complex orography on precipitation estimated by radar. Radar observations are adjusted using a network of gages. The corrections of radar estimates are derived through a Weighted Multiple Regression as a function of (1) the distance from the radar, (2) the minimum height a meteorological target must reach to be visible from the radar site, (3) the height of the ground at each pixel. Two C-band radars 140 km apart in an Alpine region are analyzed. About 60 radar-gage data pairs are available for each radar in a 25,000 km2 area during an extreme Mediterranean event (precipitation measured by the gages during the 40-hour observation period ranges from 36 to 444 mm). In the radar-gage comparison the data pairs are divided into two groups: one is used for training, the other for testing, and vice versa. All the radar-gage data pairs are used for training, when comparing the two radars. Then radar-derived precipitation estimates are compared in a mutual coverage area (≈10,000 km2). In all cases the correction significantly reduces the bias and the standard deviation of precipitation difference. In spite of the different ages, technologies, and distances from the area of interest, both radars have shown similar behavior, and the proposed procedure can be successfully applied to both. ¿ 2001 American Geophysical Union