SummaryAfter its initial discovery in Macedonia in 1985, during the last 19 years the leafminer mothCameraria ohridellahas invaded most of Central and Western Europe. The species, which causes aesthetic damage to horse chestnuts, is generally observed first in highly populated locations before colonizing the countryside. This pattern is consistent with a stratified dispersal process combining long‐distance movements and local diffusion.Using large‐scale spatial data on damage caused byCameraria ohridellain Germany, three stochastic spatial models of spread are compared: a diffusion model, a leptokurtic dispersal model and a stratified dispersal model that assumes a two‐scale dispersal process. In addition, the association between human population and moth invasion is tested in this last model by linking long‐distance infestation probability to human population density. Finally, these spatial models constructed with data from Germany are tested at the European scale and compared to historical records of first occurrence.The fat‐tailed dispersal kernel models (leptokurtic and stratified dispersal models) allowing for long‐distance dispersal provide better predictions than the diffusion model. Among these models, the stratified dispersal model incorporating the effect of human population density provides the best description of the spread ofCameraria ohridellain Germany in predictive (lowest sum of squared errors) and qualitative (similar fractal dimension) terms.The roles of short‐distance and long‐distance dispersal inCameraria ohridellainvasion ecology in relation to human population are discussed, together with the models’ scale‐dependence and limitations.