A model for photochemical evolution in power plants has been developed and used in combination with measurements to investigate ozone production efficiency (OPE) per NOx and the rate of photochemical removal of NOx. The model is a two-dimensional (2-D) Lagrangian model with 1 km horizontal resolution and vertical resolution ranging from 25 m to 300 m, nested within a larger 3-D regional model. These results are compared with measured O3, NOx, and SO2 along aircraft transects through power plant plumes (Ryerson et al., 1998). Measured flux of NOx and SO2 in plume transects is 33%--50% lower than plume emission rates, suggesting that a fraction of plume emissions remained above the top of the convective mixed layer. The measured loss rate of SO2 is greater than expected from photochemistry and surface deposition, suggesting the possibility of cumulus venting of the boundary layer. If model parameters are adjusted to reflect this hypothesized transport of plume air above the boundary layer, then the model NOx removal rate agrees with measurements. OPE is derived from measured fluxes of O3, NOx and SO2 in the plume. The resulting OPE (1.2--2.5) is somewhat higher than the previous estimate by Ryerson et al. OPE in models is slightly higher than measurements (2--3). Higher OPE (3--4) is predicted for power plant plumes 18 hours downwind of the emission source. OPE-for these far downwind plumes is comparable to OPE for other emission sources of NOx. OPE and NOx lifetime are both correlated with plume NOx concentrations. Model results suggest that OPE inferred from statistical correlations between O3 and tracers such as SO2 underestimate the true OPE in situations where the correlation between O3 and SO2 is nonlinear. ¿ 2000 American Geophysical Union