The logistic model is commonly used for analysis of discrete, multinomial data. Such a model was used for the statistical evaluation of data concerning infection of field pea varieties by downy mildew, in two series of field trials. Each series consisted of experiments performed in locations spread over the whole of Poland in the time period from 2002 to 2005. Varieties cultivated on light soils were compared in the first series, and varieties cultivated on rich soils in the second. The most resistant varieties were identified (Sokolik - light soils, Terno - rich soils) and significant differences among varieties were detected. Estimators of model parameters were found using the Fisher scoring method implemented in logistic glm procedure of the SAS system.