Introduction Many recent economic studies of the decision whether or not to adopt a new agricultural practice emphasise that the decision involves a dynamic learning process (e.g. Abadi Ghadim and Pannell 1999; Marra et al. 2003) . Perceptions associated with the relative advantage or utility of a new practice are a common focus of this approach, with several empirical studies having demonstrated the influence of such farmer perceptions of technology characteristics on adoption decisions (e.g.

more »

... desina and Baidu-Forson 1995; Cary and Wilkinson 1997) . The process where farmers revise and update their prior perceptions concerning an innovation by collecting and assimilating further information was first formalised in an economic model of the adoption process by O'Mara (1971) . In an effort to further improve the specification of the adoption process, Lindner et al. (1979) used the approach to better explain the observed time lag between when a decision maker first learns of the existence of an innovation to when a decision is made to adopt it. Within this learning-based conceptual framework for adoption, major roles for agricultural extension include facilitating (and accelerating) the learning process and potentially influencing farmer perceptions via the generation, sharing and/or delivery of effective information. To effectively target investment in research and extension it is useful to have an understanding of the major factors that influence farmers' adoption decisions in the farming system. A large number of adoption studies have been conducted with this broad objective, but the contribution of many such studies to improved planning of research and development and extension has often been disappointing. We propose that an important factor contributing to this disappointment is that relatively few adoption studies identify the influential Abstract. Early identification of farmer perceptions influencing particular farm management decisions provides the opportunity to more effectively focus investment in research and extension. A survey-based study examining the adoption of integrated weed management by Western Australian grain growers was used to demonstrate how identification of key farmer perceptions can help to guide research and extension priorities. It was found that the adoption of integrated weed management practices was influenced by grower perceptions of herbicide resistancerelated factors and of the efficacy and economic value of integrated weed management practices in the farming system. However, there were generally no significant differences between the perceptions of practice efficacy held by users and non-users of the integrated weed management practices. As initial perceptions of efficacy were generally consistent with local field experience, it was expected that extension would not have a major influence on this variable. Consistent with this, participation by growers in a workshop based on the bio-economic farming systems model, resistance integrated management (RIM), did not result in changes in perceptions of practice efficacy. However, changes in the perceived short-term economic value of some weed management practices did occur where the broader value of practices to the farming system, not necessarily relating to weed control, could be demonstrated. This also led to more growers deciding to adopt those practices. For example, intended wheat seeding rates were shown to increase by 5 kg/ha as a result of participation in the extension activity. Determining the perceptions influencing adoption, and then identifying the major learning opportunities can be valuable in focusing research and extension. Measures of perceptions also allow learning to be evaluated. In the case study of adoption of the integrated weed management practices in WA, it seems that emphasis on developing and extending the farming-systems impacts beyond just weed and resistance management is likely to be more effective than focusing on the efficacy of the practices for controlling major weeds.