Using the 11-level Geophysical Fluid Dynamics Laboratory (GFDL) global chemical transport model (GCTM) with all known sources of tropospheric NOx, we simulate the global tropospheric distribution of peroxyacetyl nitrate (PAN) and quantify its impact on tropospheric NOx. The model's global distribution of PAN is in reasonable agreement with most available observations. In the atmospheric boundary layer, PAN is concentrated over the continental sites of NOx emissions, primarily the midlatitudes in the northern hemisphere and the subtropics in the southern hemisphere. PAN is distributed relatively zonally throughout the free troposphere of the northern hemisphere, with the maximum levels found in the coldest regions, while in the southern hemisphere the maximum PAN levels are found in an equator to 30¿S belt stretching from South America to Australia. Overall, the simulated three-dimensional fields of seasonal PAN are a result of the interaction of the type of transport meteorology (convective or synoptic scale storms) occurring in the PAN formation regions and PAN's temperature-dependent lifetime. We find the impact of PAN chemistry on NOx to be rather subtle. The magnitude and the seasonal cycle of the global tropospheric integral of NOx, which has its maximum in January and the formation of HNO3 as its dominant loss path, are barely affected by the inclusion of PAN chemistry, however PAN, as a result of its temperature sensitivity and transport, regionally provides an efficient mechanism for redistributing NOx far from its source areas. With the inclusion of PAN chemistry, monthly mean NOx concentrations increase by up to a factor of 5 in the remote lower troposphere and show a spring maximum over areas of the North Atlantic and North Pacific Oceans. In contrast, PAN has only a minor impact in the upper half of the troposphere (¿10%). Examining local time series of NOx and PAN, the monthly mean mixing ratios in remote regions are shown to be composed of numerous short-term (1--2 days) large magnitude events. These episodes are large enough to potentially result in ozone production even when the monthly mean NOx values are in the ozone destruction range. While both the direct transport of NOx and its indirect transport as PAN contribute to the elevated NOx episodes over the remote extratropical oceans, events over the remote subtropical oceans are dominated by midtropospheric PAN that sinks anticyclonically equatorward and decomposes to NOx in the warmer air. ¿ American Geophysical Union 1996