Results from an ensemble of models are used to investigate the response of lightning nitrogen oxide emissions to climate change, and the consequent impacts on ozone production. Most models generate lightning using a parametrization based on cloud-top height. With this approach and a present-day global emission of 5TgN, we estimate a linear response with respect to changes in global surface temperature of +0.44±0.05TgNK−1. However, two models using alternative approaches give +0.14 and −0.55 TgNK−1 suggesting that the simulated response is highly dependent on lightning parametrization. Lightning NOx is found to have an ozone production efficiency of 6.5±4.7 times that of surface NOx sources. This wide range of efficiencies across models is partly due to the assumed vertical distribution of the lightning source and partly to the treatment of NMVOC chemistry. Careful consideration of the vertical distribution of emissions is needed, given its large influence on ozone production.