The report describes the development of a model for predicting industrial firms' choices of boiler fuel for production of process steam. In contrast to fuel choice models currently in operation, the model described here predicts fuel choices probabilistically, rather than deterministically. In addition, the model is empirical, does not assume that the costs relevant to fuel choices necessarily are best summarized by a single, total annualized cost, and provides a way to represent the effects of non-cost factors that may influence industrial boiler fuel choice decisions. These characteristics enable the model to provide a considerably more accurate description of fuel choices than do deterministic models that assume these choices to be based only on considerations of total annualized costs. The work was instrumental in developing a clearer understanding of the relative importance of the factors at work that have major influence on the industrial boiler fuel choice decision.