Context Crop rotations, within-field mixtures, and landscape mosaics including susceptible and resistant crops are three commonly adopted crop diversification strategies that can limit crop epidemics. Typically, the effects of crop diversification at these three scales have been studied separately, on single pathogen species, and with low environmental variability. Objectives We aim to compare the disease-limitation effect of these three types of crop diversification on two highly damaging fungal pathogens of wheat Puccinia recondita (WLR) and Zymoseptoria tritici (STB) and under varying weather conditions (warmer or cooler climate for WLR, wetter or drier conditions for STB). Methods We built a dynamic mathematical model of epidemics at the field scale (based on classical Susceptible-Exposed-Infectious-Removed epidemiological models) embedded in a spatially explicit landscape grid framework. We use it to simulate an agricultural landscape in which diversification translates into different proportions of wheat and resistant crops in the landscape. Results In our simulations, for both pathogens and in all weather conditions, within-field crop mixtures had the greatest impact in limiting epidemics, crop rotations were second-best, while landscape mosaics were the least effective. We also found that the threshold above which further addition of resistant plants to crop mixtures would not cause further disease limitation to be dependent on weather conditions. The more favorable the weather is for pathogens the more resistant plants are required. Conclusions Our findings imply that interactions between spatial scale of crop diversification, pathogen characteristics and weather conditions should be considered in order to maximize benefits from disease-regulation properties of diversified cropping systems under climate change.