A 'virtual hierarchy' model is described for studying the spread of pathogens between herds of livestock. This novel approach to simulating disease has animals, herds, and geographic regions in a national livestock industry arranged as a hierarchy of objects in computer memory. Superimposed on all objects is an infection-recovery cycle, a control programme, and surveillance based on test results and animal movement. The model was applied to predicting progress in the control of Salmonella Dublin in the Danish dairy cattle industry over a 10-year period. More frequent testing of bulk tank milk for antibodies to S. Dublin was less effective than improved herd biosecurity. Restricting cattle movement between regions provided a strong benefit to those regions initially with a low prevalence of infection. Enhanced control within infected herds was of intermediate benefit. A combination of strategies was highly effective although cost and feasibility of this option needs further exploration.