We have constructed profiles of lightning NOx mass distribution for use in specifying the effective lightning NOx source in global and regional chemical transport models. The profiles have been estimated for midlatitude continental, tropical continental, and tropical marine regimes based on profiles computed for individual storms in each regime. In order to construct these profiles we have developed a parameterization for lightning occurrence, lightning type, flash placement, and NOx production in a cloud-scale tracer transport model using variables computed in the two-dimensional Goddard Cumulus Ensemble (GCE) model. Wind fields from the GCE model are used to redistribute the lightning NOx throughout the duration of the storm. Our method produces reasonable results in terms of computed flash rates and NOx mixing ratios compared with observations. The profiles for each storm are computed by integrating the lightning NOx mass across the cloud model domain for each model layer at the end of the storm. The results for all three regimes show a maximum in the mass profile in the upper troposphere, usually within 2--4 km of the tropopause. Downdrafts appear to be the strongest in the simulated midlatitude continental systems, evidenced by substantial lightning NOx mass (up to 23%) in the lowest kilometer. Tropical systems, particularly those over marine areas, tended to have a greater fraction of intracloud flashes and weaker downdrafts, causing only minor amounts of NOx to remain in the boundary layer following a storm. Minima appear in the profiles typically in the 2--5 km layer. Even though a substantial portion of the NOx is produced by cloud-to-ground flashes in the lowest 6 km, at the end of the storm most of the NOx is in the upper troposphere (55--75% above 8 km) in agreement with observations. With most of the effective lightning NOx source in the upper troposphere where the NOx lifetime is several days, substantial photochemical O3 production is expected in this layer downstream of regions of deep convection containing lightning. We demonstrate that the effect on upper tropospheric NOx and O3 is substantial when the vertical distribution of the lightning NOx source in a global model is changed from uniform to being specified by our profiles. Uncertainties in a number of aspects of our parameterization are discussed. ¿ 1998 American Geophysical Union