To obtain representative relationships between rain intensity and radar reflectivity and make an accurate estimation of the rainfall amount in the Baiu season (the most important rainy season in eastern Asia), the precipitation types (convective, stratiform and mixed types) must first be classified. This is because the mechanism and efficiency of precipitation show distinct variations according to rain regimes. An objective classification method, bright band fraction (BBF) that focuses on the vertical structure of radar reflectivity field, was improved and applied to the Baiu precipitation of 1995 in the Kanto Plain, Japan. Samples of convective rainfall were further divided into patterns using an index effective efficiency (Ee) that represents the echo top height. For those types (and patterns) that had been divided, Z-R relations were established using a statistical method called the window probability matching method (WPMM). The shapes of the Z-R relations show a natural matching of the probability density function of radar reflectivity with that of rain intensity recorded by rain gauge. The results show that the classification of precipitation types is reasonable, and Z-R relations corresponding to the precipitation types are representative. Some facts which need special attention in the application of the WPMM are also revealed.