Fossil fuel combustion accounts for >50% of the global atmospheric emission of NOx, but this source is concentrated in the polluted continental boundary layer (CBL) and only a small fraction is exported as NOy (NOx and its oxidation products) to the global troposphere. Better quantification of this export efficiency is needed because of its implications for global tropospheric ozone. A recent Lagrangian analysis of the NOy-CO correlations observed from the North Atlantic Regional Experiment in September 1997 (NARE'97) aircraft campaign downwind of eastern North America (September 1997) indicated a NOy export efficiency of <10%, with <10% of the exported NOy present as NOx. In contrast, previous three-dimensional (3-D) model Eulerian budget analyses for the North American boundary layer indicated NOy export efficiencies of 25--30%, with 30--35% of the exported NOy present as NOx. We investigated this apparent discrepancy by simulating the NARE'97 aircraft observations with a global 3-D model of tropospheric chemistry (GEOS-CHEM) and using the model to calculate the NOy export efficiency both through a Lagrangian analysis of the NOy-CO correlations along the aircraft flight tracks and through an Eulerian budget analysis for the North American boundary layer. The model reproduces the variability and NOy-CO correlations observed in the aircraft data and also at the Harvard Forest surface site in the northeastern United States. We show that the previous Lagrangian analyses of the NOy export efficiency during NARE'97 were probably biased low because of underestimation of the CO background. Correcting for this bias, we find a NOy export efficiency of 17 ¿ 7% in the model and 15 ¿ 11% in the observations. A similar NOy export efficiency (20%) in the model is obtained from the Eulerian budget analysis, demonstrating that the Lagrangian and Eulerian approaches are in fact consistent. Export efficiencies of NOy in previous 3-D model Eulerian budget analyses were probably too high because of insufficient scavenging out of the CBL. Model results indicate that only 6% of the exported NOy is present as NOx along the aircraft flight tracks, in agreement with the observations, but that 40% of the NOy export flux is present as NOx, in agreement with the previous 3-D model analyses. This result reflects the fast oxidation of NOx between the point of exit from the CBL and the point of sampling by the aircraft. The eventual ozone production in the global troposphere due to exported NOx and peroxyacetylnitrate (PAN), with equal contributions from each, is comparable in magnitude to the direct export of ozone pollution from the North American boundary layer.