Cucumber mosaic virus (CMV) causes a serious disease of narrow-leafed lupin (Lupinus angustifolius). It is seed-borne in lupin and seed-infected plants act as the primary virus source for secondary spread by aphid vectors within crops. Infection with CMV causes yield losses of up to 60% in epidemic years, but has little impact on yield in years when spread is limited. Aphids also cause sporadic yield losses due to direct feeding damage. A simulation model was developed to forecast aphid outbreaks and epidemics of CMV in lupin crops growing in the 'grainbelt' of south-west Australia, which has a Mediterranean-type climate. The model uses rainfall during summer and early autumn to calculate an index of aphid build-up on weeds, crop volunteers and self-regenerating annual pastures in each 'grainbelt' locality before the growing season commences in late autumn. The index is used to forecast the timing of aphid immigration into crops. The subsequent aphid build-up and movement within the crop, spread of CMV from virus-infected source plants within the crop, yield losses and percentage of harvested seed-infected are then calculated. The model evaluates the effects of different sowing dates, percentages of CMV infection in seed sown and plant population densities on virus spread. The model simulations were validated with 14 years' field data from six different sites in the 'grainbelt', representing a wide range of pre-growing season rainfall scenarios, sowing dates, percentages of infection in seed sown and plant population densities. The model was incorporated into a decision support system (DSS) for use by lupin farmers and agricultural consultants in planning CMV management and targeting sprays against aphids to prevent direct feeding damage. The inputs required from the user are lupin cultivar, anticipated emergence date, percentage CMV infection in seed sown, plant density and location. The output consists of a personalised risk forecast for the user and includes predictions for date of first aphid arrival, aphid numbers, CMV spread, final virus incidence, yield loss due to infection and percentage infection in harvested seed. Predictions from the DSS are accessible via an Internet site and from other information sources. The model can serve as a template for modelling similar virus/aphid vector pathosystems in other regions of the world, especially those with Mediterranean-type climates.