Natural resource managers often face hard decisions with uncertain outcomes. Common questions we have encountered include: What is the best way to conserve fish populations in the presence of dams when other threats are also present? or Will removing this migration barrier to benefit native fish let non-native fish invade? In these cases, empirical data may not be available or the question may be very complex or have many potential outcomes. Decision tools, like Bayesian networks, can help in these situations because they help organize thinking and information, provide a transparent record of the logic that led to a decision, and identify the important data gaps that, if filled, would improve future decision making. We have built decision analysis models for a variety of purposes, including predicting habitat suitability under, exploring management alternatives for a threatened fish species, and making individual decisions about whether to remove or harden a barrier.