Ozone pollution episodes and heat waves have negative impacts on human health and can damage vegetation. These two types of event are linked by their occurrence during stagnant conditions and the influence of temperature on emissions of biogenic precursors to ozone formation. Subseasonal forecasting skill and predictions of the frequency and intensity of these extreme events under climate change could be improved by characterizing large-scale meteorological patterns (LSMP) that typically occur in advance. Ozone episodes and heat waves are identified in records from North American ozone and temperature monitoring stations. Statistical clustering methods are used to divide stations into large regions (e.g. West Coast, Northeast United States) according to their likelihood to simultaneously experience extreme temperatures or ozone concentrations. These divisions are useful because the characteristic LSMPs for extreme events differ by region. Upper troposphere composite circulations are calculated from reanalysis data for heat waves and ozone episodes in each region. Wavenumber-frequency analysis is used to decompose these patterns into standing and travelling zonal wave components. Persistent planetary-scale standing wave patterns are found to precede both ozone episodes and heat waves in most regions. It is argued that because these patterns evolve more slowly than typical weather patterns, they may be used to better predict future ozone and temperature extremes.