Because the analysis of risky choice in agriculture and rural resource management is important but difficult, we argue that there is a need for some agreed principles on how to proceed. This paper is intended as a first step to this end. We start with the proposition that the importance of risk aversion has generally been exaggerated relative to the task of finding better ways to deduce relevant and reliable probabilities. Getting better probabilities demands careful thought, drawing on what is known about the pitfalls and on evolving insights into better ways of proceeding. Our aim is to stimulate a debate leading to a clearer consensus about better practice in these matters.