This article analyses the role of research and development (R&D) in Australia's broadacre farming by using the semi-parametric smooth coefficient model proposed by Hastie and Tibshirani (1993) and Li et al (2002) and a state-level dataset covering the period 1995 to 2007. While the conventional production function approach only captures the direct effects of R&D, this methodology captures both the direct impact of a change in R&D on output and the indirect impact through changes in efficiency of use of factor inputs in the production process. The empirical results show that once both the direct and indirect effects are taken into consideration, R&D investments significantly increase outputs. The results also show that there are substantial variations in the effects of R&D on output across the states. Such variations need to be taken into account when designing policies for investing public R&D in agriculture.