Chambers and Quiggin (2000) have used state-contingent production theory to establish important results concerning economic behaviour in the presence of uncertainty, including problems of consumer choice, the theory of the firm, and principal-agent relationships. Empirical application of the state contingent approach has proved difficult, not least because most of the data needed for applying standard econometric methods are lost in unrealized states of the world. O'Donnell and Griffiths (2006) show how a restrictive type of state-contingent technology can be estimated in a finite mixtures framework. This paper shows how Bayesian methodology can be used to estimate more flexible types of state-contingent technologies.