Abstract: Using a 40-year daily precipitation dataset including 134 stations from 1962 to 2001, the large-scale distribution patterns of precipitation anomalies over China were investigated. In the phase space spanned by the first 8 EOFs generated from the 20-day low-pass filtered data, the six rainfall regimes (RRs) were id entified by applying a cluster analysis method, namely, the northeastern China regime, Yellow River regime, Qinling-Huaihe regime, Yangtze River with its south regime, Southern China regime, and rainless regime. Analyses show that the new RRs exhibit good persistence and evident physical sense, and excellently represent both of countrywide and regional features, which also demonstrate the inhomogene ity of multi-dimensional phase space. Furthermore, it is more important that th e new RRs can describe intraseasonal dynamic characteristics of large-scale rainfall anomalies,which is the most significant difference between the new RRs and the conventional season average rainfall patterns. On the other hand, the clim atic characteristics of daily distributions of the RRs events, as well as the 40-year panorama of the RRs occurring were also investigated, which further document reasonableness and objectivity of the RRs with intraseasonal variabil ity, and are likely to present more helpful information for short-range climate forecast, compared with other previous classical rainfall patterns.