CONTEXT: Coffee leaf rust (CLR) epidemics on Coffea arabica have led to severe socio-economic crises in Latin America starting in 2008. Until now, the scattered nature of scientific and empirical knowledge of the highly complex CLR-coffee pathosystem has been an obstacle to the development of CLR forecasting models. OBJECTIVE: To help prevent new severe epidemics, we built ExpeRoya, a qualitative model, based on a review of the scientific literature and expert opinion, to forecast the risk of a monthly increase in the incidence of CLR at plot and landscape levels. METHODS: We adopted the IPSIM (Injury Profile SIMulator) framework, a qualitative and aggregative modeling approach that describes the effects of the cropping system and the plot environment on injuries, thereby making it possible to incorporate scattered knowledge on the system and all its complexity in a simplified way. Involving experts makes this approach powerful and robust because it builds on empirical knowledge based on a very large number of field observations. We argue that broad expert knowledge provides more accurate information on the manifold interactions in the system than existing quantitative models can. The structure of ExpeRoya was discussed with coffee sector experts in 19 workshops and validated in an online survey with 17 CLR experts. RESULTS AND CONCLUSIONS: ExpeRoya successfully integrates in a simple way 229 multiple interactions that exist within the CLR-coffee pathosystem based on only 12 input variables easily acquired in the field: one incidence monitoring variable; two meteorological variables (temperature and rainfall), four crop management variables (management of shade cover, fungicide application, nutrition and pruning of coffee trees) and five coffee tree characteristics (dates of flowering, beginning and end of harvest, fruit load and cultivar genetic resistance). Coffee institutes in Honduras and Nicaragua now use ExpeRoya, hosted by the platform Pergamino (https://www.