The importance of snowmelt in the western US provides a strong link between cold season precipitation and warm season hydrology, and this link allows substantial predictability. Here we focus on regional-scale patterns, which are well-captured in the available data for at least 50 years. Drought is considered in terms of river flow station data, gridded Palmer Drought Severity Index (PDSI), and vegetation health as estimated by NDVI. Both simple area-average indices and Canonical Correlation Analysis (CCA) are used to extract the regional-scale patterns of covariability between cold season precipitation and subsequent warm season drought. Focusing on regional-scale patterns avoids the difficulties associated with trying to estimate snow water equivalent at sufficiently high resolution to accurately characterize snowmelt. Moreover, the regional-scale patterns are still highly correlated to individual river flow stations and 8km NDVI. The practical predictability associated with these relationships is assessed via cross-validation. The spatial structure of the regional patterns is also investigated, especially for NDVI, which shows much more spatial structure than the hydrological variables. Finally, the potential impact of warming trends is estimated by comparing the relationships among the different drought variables after the warmest 5 winters with the average relationships.