The invention discloses a method for predicting occurrence of pecan trunk rot based on a discriminant analysis method, which comprises the following steps: selecting a pecan tree sample, collecting the soil sample, and acquiring the pH value of the soil in the dormant period and the ratio of the alkaline hydrolysis nitrogen content to the quick-acting potassium content; Recording the morbidity ofthe dry rot in the next year; Establishing a discriminant analysis model by taking morbidity and non-morbidity as discriminant types, soil data as a training set and the ratio of the pH value to the alkaline hydrolysis nitrogen content to the quick-acting potassium content as model parameters; Collecting the pH of the soil of the to-be-predicted walnut tree and the data of the ratio of the alkaline hydrolysis nitrogen content to the quick-acting potassium content, importing the data into the model, and predicting whether the walnut tree suffers from the dry rot or not in the next year. The method depends on a discriminant analysis method to find index parameters which have relatively large correlation with the pecan trunk rot, namely the ratio of the pH value to the alkaline hydrolysis nitrogen content to the quick-acting potassium content, and a prediction model is obtained by utilizing the two parameters, so that the prediction accuracy rate reaches more than 80%.