Winter wheat cropping has changed considerably over the years 1974-1986 in The Netherlands. Yield has been increased from 5 to 8 ton/ha, due to short strawed cultivars and higher levels of agrochemical inputs. The changes were described. Epidemics and damage relations of powdery mildew were studied, to construct an advisory system.In surveyed fields, mildew intensity in spring was correlated with mean temperature in October and with the average temperature during December, January, February and March and with the susceptibility of the cultivars grown. The development of mildew during May-July was correlated with the area sown with a cultivar and with its susceptibility. The high variation in mildew intensity between fields points to the need of disease monitoring in management systems.To assess mildew intensity, pustule counts were made and the fraction of diseased leaves (I, incidence) was determined (Neth. J. Pl. Path. 92: 197-222). In fields, mean pustule number per leaf m, may be estimated from I by: ln(m) - 1.48 +1.14 ln(ln[l/(l-I)]). Errors of estimates were studied.Damage by mildew was studied in 13 field experiments (Neth. J. Pl. Path. 94: 69-80 and 95: 85-105). The mildew profile in the canopy was described by a model with one parameter, for which an estimate was given. Damage was described by the function: -0.013 (SE - 0.003) kg/are per pustule-day Of mildew per leaf, from second node stage to early dough at yield levels of 70 to 90 kg/are. After 1980, farmers were advised to control mildew based on data from these experiments and from the surveys.To upgrade this advisory system a simulation approach was used to explore effects of weather on yield and damage (Neth. J. Pl. Path., in press). Yields simulated were compared with yields harvested. The difference in performance of two models could be attributed to the simulation of early growth. Mildew was introduced in a model by quantification of five parameters, taking the spatial distribution of mildew into account. Computed damage approached but ...