Even long term measurements of substantially concurrent rain rate and rain attenuation pdfs, as much reliable as is experimentally possible, need to be inspected by a specialist to avoid intervals of dubious data (see also [2]). This hard, difficult and time-consuming work is the inevitable price to pay for obtaining, at last, a useful data bank of concurrent rain rate and rain attenuation statistic from the actual ITU-R data bank used to test prediction models. This was done in this paper for the data collected at Neyagawa (Japan). In the study case here discussed, we have found that the probability formulation of the synthetic storm technique [4, 6], whose parameter m (thetas, f ) has been established from Italian rain rate measurements [6], is the most accurate prediction model of rain attenuation from rain rate probability distributions.