Wintertime cold air outbreaks (CAOs) and warm waves (WWs) strongly impact the continental United States.  We analyze the behavior of US CAOs and WWs during the past 60 winters using atmospheric observations and parallel CMIP5 simulations.  No significant long-term trends in either WWs or CAOs are observed over the continental United States.  Winter-mean CAO frequency is modulated by the (i) North Atlantic Oscillation (NAO) over the U.S. Southeast and (ii) Pacific–North American (PNA) pattern over the Northwest.  WW frequency is modulated by the (i) NAO over the eastern United States and (ii) combined influence of the PNA, Pacific decadal oscillation (PDO), and ENSO over the southern United States. Linear statistical analysis reveals that the collective influence of low-frequency modes can account for as much as 50% of the year-to-year regional variability.  Although like behavior occurs in CMIP5 models, WW (CAO) frequency is typically overestimated (underestimated) and most models are unable to replicate observed linkages to the PDO. Our results suggest that predictions of future CAO and WW behavior are limited by the ability of climate models to accurately represent the primary natural modes of low-frequency variability.