Abstract. Lightning production is described using thermodynamic hypotheses mainly according to convective activity in the atmosphere. However, existing formal thermodynamic descriptions are unable to fully explain the profound difference in the flash rate between tropical Africa and South America and between land and ocean. Aerosols are shown to be regulators of lightning in regional studies, but their influence on lightning production at the global scale is not described. I analyzed spatial patterns of the satellite annual global flash rate and simulated, annually averaged cloud condensation nuclei (CCN) distribution and found consistent positive correlation between them for land, ocean and continents. I developed a simple model of lightning production that is based solely on an aerosol hypothesis. The central premise of this model is that concentration of graupel pellets and concentration of ice crystals, which both determine flash rate, are monotonically increasing by CCN concentration up to a critical value (around 2000 cm⁻³). However, ice crystal concentration falls rapidly after the threshold due to lowering in the number of large cloud droplets effective for rime-splintering ice multiplication. Comparison of the model with a model of the flash rate based on thermodynamic hypotheses demonstrates that the aerosol hypothesis explains the global annual spatial distribution of lightning production consistently better over land and over oceans. My results emphasize importance of aerosols for lightning production and point to the existence of a global aerosol-lightning feedback, which affects both the climate system and the land surface.