Johne's disease is not known to occur in Western Australia. For the sheep industry, the most significant risk comes from the importation of sheep from interstate. The risk of Johne's disease not being detected in sheep imported from New South Wales was assessed quantitatively by using a stochastic simulation model. Results from the computer model were used to support decisions about controls on the importation of sheep. The process of importing sheep was broken down into steps and numbers or probabilities assigned to each. Controls on the movement of sheep included surveillance tests in source flocks and serological tests on sheep in consignments before and after transportation to Western Australia. The model calculated the risk of occurrence of Johne's disease in Western Australia and the success of the agar gel immunodiffusion test in identifying consignments with infected sheep. Negative surveillance tests in source flocks reduced the risk to about one twentieth of that when no surveillance tests were required. On average, Johne's disease was predicted to be introduced once in every 3 to 7 years when no testing of either the source flock or the sheep in consignments was required. When only negative surveillance tests were required the interval increased to once in every 63 to 111 years and, with the additional requirement that all sheep in each consignment must have a negative test before and after transport, the interval further increased to once in every 125 to 333 years. When only sheep in consignments were tested, the interval was calculated to be 8 to 14 years. Controls on the importation of sheep were changed to provide an acceptable level of risk for the sheep industry in Western Australia.