Aiming at the uneven spatial and temporal distribution of precipitation and frequent occurrence of drought and waterlogging disasters in Guangxi, the multi-scale characteristics and comprehensive evaluation of precipitation heterogeneity can provide scientific support for regional response to drought and waterlogging disasters and intelligent management of water resources. Based on the daily precipitation data of 87 grid points from 1961 to 2017 in Guangxi, the Precipitation Concentration Degree (PCD) and Precipitation Concentration Period (PCP) index were used to build day, pentad, ten days, month, season precipitation heterogeneity of multiple time scale level evaluation system. By using R/S analysis and geographical spatial analysis methods, the space-time evolution characteristics and climate division in Guangxi were discussed. The study showed that the spatial and temporal variation characteristics of PCD and PCP at the diurnal, synoptic and monthly scales were relatively consistent, and the heterogeneity of precipitation in Guangxi could be better expressed than that at the monthly and seasonal scales. The diachronic change of PCD in precipitation in Guangxi shows an increasing trend in northeast China and a decreasing trend in southwest China, and the trend of its future period is the same as the diachronic evolution. The spatial distribution of PCD in Guangxi has significant spatial autocorrelation and stratification heterogeneity, which are mainly reflected in the mean value, coefficient of variation and frequency of PCD. The comprehensive heterogeneity of precipitation in Guangxi is highly dispersed in the northeast, highly concentrated in the south, and slightly concentrated or dispersed in the northwest and central regions. The time-scale within a month is the best scale to express the non-uniformity of precipitation in Guangxi. If the advantages of more stable climatic and ten-day scales and more fine daily scales are taken into account, the use of climatic scale for daily sliding calculation and analysis will be the best way.