Improperly designed road-stream crossings can fragment stream networks by restricting or preventing aquatic organism passage. These crossings may also be more vulnerable to high flow events, putting critical human infrastructure at risk. Climate change, which will require access to suitable habitat for species persistence, and is also predicted to increase the frequency and magnitude of extreme floods, underscores the importance of maintaining stream connectivity and resilient infrastructure. Given the large number of road-stream crossings and the expense of replacement, it is increasingly important to prioritize removals and account for the multiple benefits of these management actions. I developed an aquatic barrier prioritization scheme that combines potential habitat gain, stream thermal resilience, aquatic organism passage, and culvert risk of failure. To assess relative thermal resilience, I deployed paired air-water thermographs in several New England watersheds and analyzed relative thermal sensitivity (relationship of water to air temperature) and exposure (duration, frequency, and magnitude of warm stream temperature episodes) among streams. These were combined into a single metric of thermal resilience corresponding with the distance of that stream’s sensitivity and exposure from the watershed median. To test the relationship between risk of failure, culvert dimensions, and stream connectivity, I developed a logistic regression to predict risk of failure using data from two watersheds that experienced extreme flooding from Hurricane Irene (2011). Finally, I applied the resultant prioritization scheme to 66 road-stream crossings in the Westfield River watershed (MA). Thermal habitat quality varied considerably within and among watersheds. Stream sensitivity was generally lower than the widely accepted 0.8 ̊C increase in stream temperature for every 1 ̊C increase in air temperature (Westfield median sensitivity = 0.44), with substantial differences among streams. Exposure also varied widely among streams, indicating that some headwater streams in New England are more thermally resilient than previously thought. Risk of infrastructure failure was predicted with a logistic regression using culvert constriction ratio and predicted aquatic organism passage as predictors (Likelihood ratio test, X2=59.1, df=3, p- value=9.2e-13), indicating that underdesigned culverts were more likely to be barriers to passage and more likely to fail in extreme flow events. To prioritize culverts, this study ultimately used a piecewise approach that identified culverts opening the longest reaches of thermally resilient habitat, and then ranked those culverts by infrastructure replacement need. In the Westfield River, the prioritization clearly identified crossing replacements most likely to yield multiple benefits. The scheme I developed can accommodate changes in the relative weights of the different criteria, which will reflect differences in management and conservation concerns in the confidence of inputs. In conclusion, increasing connectivity by removing barriers may be one of the most effective ways to mitigate the effects of climate change on aquatic systems, but it is important to remove the right barriers.