A new dynamical metric of troposphere-stratosphere coupling is established based on extreme stratospheric planetary-scale wave heat flux events, defined as the 10th and 90th percentiles of the daily high-latitude averaged heat flux distribution at 50 hPa using European Centre for Medium-Range Weather Forecasts Re-Analysis (ERA)-Interim reanalysis data. The stratospheric heat flux extremes are linked instantaneously to high-latitude planetary-scale wave patterns in the troposphere and zonal wind, temperature, and mean sea level pressure anomalies in the Atlantic basin. The impacts are reminiscent of different phases of the North Atlantic Oscillation. In particular, extreme positive (negative) heat flux events in the stratosphere are associated with an equatorward (poleward) jet shift in the North Atlantic basin. The metric is used to evaluate troposphere-stratosphere coupling in models participating in the Coupled Model Intercomparison Project Phase 5 (CMIP5). The results show that models with a degraded representation of stratospheric extremes exhibit robust biases in the troposphere relative to ERA-Interim. In particular, models with biased stratospheric extremes exhibit a biased climatological stationary wave pattern and Atlantic jet stream position in the troposphere. In addition, these models exhibit biases in geopotential height and zonal wind extremes in the North Atlantic region. The stratospheric biases are connected to model lid height, but it is not sufficient for assessing the tropospheric impacts. Our analysis reveals that the mean bias of the stratospheric heat flux is also not sufficient for assessing the representation of troposphere-stratosphere coupling. Overall, the results suggest that a metric based on stratospheric heat flux extremes should be used in conjunction with metrics based on extreme polar vortex events in multimodel assessments of troposphere-stratosphere coupling.