Earthworms affect various soil ecosystem processes in their role as ecosystem engineers. The spatial distribution of earthworms determines the spatial distribution of their functional effects. In particular, earthworm-induced macropore networks may act as preferential flow pathways. In this research we aimed to determine earthworm distributions at the catchment scale with species distribution models (SDMs). We used land-use types, temporally invariant topography-related variables and plot-scale soil characteristics such as pH and organic matter content. We used data from spring 2013 to estimate probability distributions of the occurrence of ten earthworm species. To assess the robustness of these models, we tested temporal transferability by evaluating the accuracy of predictions from the models derived for the spring data with the predictions from data of two other field surveys in autumn 2012 and 2013. In addition, we compared the performance of SDMs based (i) on temporally varying plot-scale predictor variables with (ii) those based on temporally invariant catchment-scale predictors. Models based on catchment-scale predictors, especially land use and slope, experience a small loss of predictive performance only compared with plot-scale SDMs but have greater temporal transferability. Earthworm distribution maps derived from this kind of SDM are a prerequisite for understanding the spatial distribution patterns of functional effects related to earthworms. Highlights: Do plot-scale variables explain earthworm distributions better than landscape-scale variables? We modelled ten earthworm species and give comprehensive model evaluations. Landscape-scale variables are almost as useful as proximate plot-scale variables. Landscape-scale variables allow temporally robust mapping, but are limited to the study area.