Abstract Rivers and streams worldwide have degraded ecosystems due to a wide range of anthropomorphic drivers and pressures. Recent decades have seen the rise of an interdisciplinary science of stream restoration, which seeks to rehabilitate damaged systems to yield degraded or destroyed ecosystem services. About $1 billion is expended annually on such projects in the United States alone, exclusive of several extraordinary projects that target whole regions or long reaches of larger rivers. The best approaches for restoring rivers are far from settled, as few projects have been sufficiently monitored to positively document effects. Future efforts should feature monitoring and sharing of information, working at larger spatial scales, and restoration of key processes rather than forms.