Assisted colonization (AC), translocating a species outside its indigenous range to avoid its extinction, is one of the few conservation options for some species. It is also controversial because of the history of ecological impacts of invasive species, including the extinction of native species as a result of novel ecological interactions resulting from the introduction. Although several national and international organizations have issued guidelines related to AC, none allow case-specific decision-making based on risks and benefits to biodiversity. We propose a two-pronged approach to fill this gap. The first step aims to separate clear-cut cases of AC from those that require an in-depth risk analysis. We propose a set of seven qualitative criteria to identify AC projects that are clearly low-risk and high-benefit, and therefore should not be controversial, and those that are clearly high-risk or low-benefit and therefore should not be attempted. This identifies only the most obvious cases, leaving out many cases to be determined through a quantitative analysis to estimate the probabilities of extirpation of the resident species because of AC, which is the second step of our approach. We propose a roadmap for developing such a system based on community ecology theory, and a framework for considering the estimated probabilities in a global context. Our framework recommends an AC project only if it would result in a larger number of globally extant species than a scenario of no action. We propose large-scale testing of the clear-cut approach, further development of the quantitative approach, and wide consultation for adopting international guidelines for risk assessment of AC projects.