Application of the window probability matching method to radar and rain gauge data that have been objectivelyclassified into different rain types resulted in distinctly different Ze-R relationships for the various classifications.The classification parameters, in addition to the range from the radar, are (a) the horizontal radial reflectivitygradients [dB km-1; (b) the cloud depth, as scaled by the effective efficiency; (c) the brightband fraction withinthe radar field window; and (d) the height of the freezing level. Combining physical parameters to identify thetype of precipitation and statistical relations most appropriate to the precipitation types results in considerableimprovement of both point and areal rainfall measurements. A limiting factor in the assessment ofthe improvedaccuracy is the inherent variance between the true rain intensity at the radar measured volume and the rainintensity at the mouth of the rain gauge. Therefore, a very dense rain gauge network is required to validate mostof the suggested realized improvement. A rather small sample size is required to achieve a stable Ze-R relationship (standard deviation of 15% of Rfor a given Ze)-about 200 mm of rainfall accumulated in all gauges combined for each classification.