Austrocedrus chilensis forests suffer from a disease caused by Phytophthora austrocedrae, which is found often in wet soils. We applied three widely used modelling techniques, with different data requirements, to model disease potential distribution under current environmental conditions: Mahalanobis distance, Maxent and Logistic regression. Each model was built using field data of health condition and landscape layers of environmental conditions (distance to streams, slope, aspect, elevation, mean annual precipitation and soil pH NaF). We compared model predictions by area under the receiver operating characteristic curve and Kappa statistics. A reasonable ability to predict observed disease distribution was found for each of the three modelling techniques. However, Maxent and Logistic regression presented the best predictive performance, with significant differences with respect to the Mahalanobis distance model. Our results suggested that if good absence data are available, Logistic regression should be used in order to better discriminate sites with high risk of disease. On the other hand, if absence data are not available or doubtful, Maxent could be a very good option. The three models predicted that around 50% (49-56%) of the currently asymptomatic forests are located on sites at risk of disease according to abiotic factors. Most of these asymptomatic forests surround the current diseased patches, at distances lower than 100 m from diseased patches. Management considerations and the scope of future studies were discussed in this article.