||We propose using regression trees as a flexible and intuitive tool for modelling the relationship between weather conditions and day to day changes of the visitor load in outdoor recreation areas. Regression trees offer a number of advantages when compared e.g. to linear models, specifically by outlining different seasonal and meteorological scenarios. When applied to video monitoring data from the Lobau, an Austrian nature conservation area, good regression tree models for the total number of visitors and the counts for some visitor categories (bikers, hikers, swimmers) were found, while other categories could not be adequately represented (dog walkers, joggers). The regression trees indicate a strong relationship between weather and total visitor numbers, as well as weather and the number of bikes and swimmers, respectively. The relationship to weather was found to be only slight for hikers and dog walkers, and completely absent for joggers. In general, the use of derived meteorological quantities in form of thermic comfort indices for characterizing weather conditions results in better models than the use of directly observable meteorological quantities.