The ongoing changes in climate are threatening human and natural systems by intensifying hydrometeorological extreme events. The purpose of this study, therefore, is to promote our ability to coherently understand and predict simultaneous extreme rainfall and drought events across the continental US by identifying their spatiotemporal distributions and inferring the significant drivers that govern their variabilities. Precipitation data from 1244 Global Historical Climatology Network (GHCN) stations, which are the stations subjected to a certain level of quality assurance, are used for generating an outlier-indicating sparse matrix based on robust principal component analysis. The sparse-matrix is investigated for spatial coherence of the joint extreme events for each year. Finally, a hierarchical multinomial regression model for these outliers is built to jointly explain the space-time variance of extreme rainfall and droughts for each station using climatic oscillations (i.e., ENSO, PDO, NAO). This regional-to-global scale analysis of simultaneous extremes is expected to advance predictive models of hydrometeorological extremes in terms of accuracy, and provide a better understanding of the coherence between extreme rainfall and droughts.