Daily circulation (sea level pressure) maps over east Atlantic-Europe were classified into six circulation patterns (CPs) by using k-means clustering coupled with principal component analysis. The relationships between winter monthly rainfall in Portugal and the occurrence frequencies of the CPs were also examined. The CP classification scheme is effective not only in discriminating regional wet/dry weather conditions, but also in differentiating between various conditions associated with different precipitation intensities. For southern Portugal, CP6 and CP3 are the wettest patterns: under their presence the rainfall occurrence probability and the rainy day rainfall intensity are twice as high as those associated with the dry CPs; with only about 26.6% occurrence probability the two wet CPs together contribute about 66.8% of the total rainfall. For northern Portugal, CP6 and CP4 turned out to be the wettest patterns. They produce over 60% of the total rainfall with only 31% chance of occurrence. In terms of long-term variation the precipitation is also intimately related to the occurrence frequencies of the CPs. The occurrence frequencies of CPs were found to be good predictors for the regional precipitation. In addition, it is evident that the decrease of March rainfall throughout the country in the last 3 decades has been a consequence of a reduction in the occurrence frequency of the wettest circulation pattern, CP6, which coincides with the upward trend of the North Atlantic Oscillation index and could be related to the northeastward shifting and strengthening of the Azores High.¿ 1997 American Geophysical Union