Airborne observations of NO during the Subsonics Assessment Ozone and Nitrogen Oxides Experiment (SONEX) reveal episodes of high NOx in the upper troposphere believed to be associated with lightning. Linkage to specific periods of lightning activity is possible through back trajectories and data from the National Lightning Detection Network. Lagrangian model calculations are used to explore the evolution of these high NOx plumes over the 1-2 days between their introduction and subsequent sampling by NASA's DC-8 aircraft. Simulations include expected changes in HNO3, H2O2, CH3OOH, HO2, and OH. Depending on the time of injection and dilution rate, initial NOx concentrations are estimated to range from 1 to 7 ppbv. Similar to many previous studies, simulated HNO3 concentrations tend to be greater than observations. Several possible explanations for this difference are explored. H2O2 observations are shown to be consistent with removal in convective activity. While it is possible that upper tropospheric CH3OOH is enhanced by convection, simulations show such increases in CH3OOH can be short-lived (e.g., < 12 hours) with no perceptible trace remaining at the time of sampling. High NO levels further prevent elevated levels of CH3OOH from propagating into increases in H2O2. HO2 is suppressed through reaction with NO in all cases. Simulated increases in OH exceeded a factor of 2 for some cases, but for the highest NOx levels, loss of OH via OH+NO2 offset production from HO2+NO. Additional increases in OH of 30-60% could result from convection of CH3OOH. A final point of discussion concerns how the chemistry within these plumes, their long-range transport, and their potential importance in sustaining background NOx far from source regions present a challenge to global and regional model simulations. Copyright 2000 by the American Geophysical Union.