in the Netherlands, from 1950 to 1996, was analysed using agronomic and meteorological variables. The intensity of annual epidemics was characterized by an index of disease intensity (DI, 0 = absence of late blight; 4 = severe epidemic). Three periods were identified, with average DIs of 2?9, 0 and 2, respectively. Period I (1950-68) had relatively regular epidemics; period II (1969-78) was virtually blight free; and period III (1979-96) showed large variations in disease intensity. Disease-enhancing factors were number of days with precipitation, and number of hours with temperatures between 10 and 27°C and relative humidity >90during the growing season. Limiting factors were number of hours with temperatures >27°C, and amount of global radiation. Linear discriminant analysis of DI using the blight status of the previous year and meteorological variables correctly classified up to 40 years out of 47 (87?0Œ with five out of the six incorrectly classified years falling in period III. Blight status of the previous year and number of days with precipitation were important discriminating variables.