The spillover effects of agricultural research are of interest because research-induced supply shifts in non-target regions can reduce the benefits for producers in the target regions. The introduction of semi-dwarfing genes in wheat into Australia provides an example of spillover from the CIMMYT program. Australia's wheat-growing environments were not those for which the CIMMYT material was specifically targeted. However, some of these lines were introduced into Australia and used in breeding programs to produce important supply shifts in Australia. An examination is made of the effects that the spillover had in Australia. Using an index of varietal improvement, with its attendant qualifications, an estimate of the extent of that shift resulting from the CIMMYT -based varieties was obtained. While the extent varied widely between states, the shift was found to be between 0.2 and 7.7%, with an overall average for Australia of 3.5% by 1983. On the basis of an assumption of perfectly elastic export demand for Australian wheat, the estimated total cost savings to Australian producers were US$747 million (in 1983-84 dollars) for the period 1974 to 1983, or an average ofUS$75 million per year. The annual contribution of Australia to CIMMYT has averaged approximately US$340,000 in recent years, while the average annual expenditure on wheat breeding in Australia has been US$4 to 5 million. On the basis of pedigrees, approximately two-thirds of the cost savings of CIMMYTbased varieties could be attributed to CIMMYT per se, with the remaining one-third attributable to the inputs of the Australian wheat breeders.