The relationship between severe winter weather in the United States and global climate indices is investigated using data from 1950 to 2002. Winter severity is characterized by accumulated freezing degree-days (AFDD) and damaging freezing rain storms. The first eigenvector of the AFDD anomaly, based on data from 2282 weather stations, shows a pattern in which the entire country tends to be either warmer or colder than average, while the second eigenvector has opposite deviations in the East and West. The relationship between the first three principal component time series and five climate indices, PDO, SOI, PNA, NAO, and Northern Hemisphere temperatures, is investigated using step-wise linear regression. This analysis shows that the most important indices for explaining the annual variation in AFDD are the NAO and PNA. An estimate of the AFDD anomaly for the winter of 2003 using analog years to hindcast the climate indices significantly underestimates the warmth in the West and the cold in the Northeast. The annual variation in the area and path of ice storms is only weakly correlated with the NAO. However, the averages of the SOI and NAO in winter months with ice storms are closer to zero than their averages in winter months without ice storms. This investigation will be extended to other regions of the world to characterize the regional AFDD as well as summer temperature patterns and determine their relationships to global climate cycles