Expected utility (EU) theory is used to derive a simple normative model for ranking endangered species preservation programs. In contrast to the current priority system, the EU model requires that potential increases in survival probabilities be combined with utility attributes (eg, uniqueness) multiplicatively. Five studies assessed the descriptive adequacy of the EU model and the importance of various utility attributes. When little information was provided, participants' concern for species was highly correlated with species' phylogenetic status, rated similarity to humans, and rated ecological importance. When more complete information was provided, participants' use of attributes related to similarity decreased. Participants valued species according to their ecological roles, uniqueness, usefulness to humans, and intelligence, but combined this information with probabilities in an inappropriate manner. In fact, many participants distinguished among species on the basis of utility attributes even when preservation programs were completely ineffective (ie, when the potential probability increase was 0). Recommendations for policy and research are offered in light of the model's descriptive performance.