Recent dramatic increases in prices for medical liability insurance, directors and officers insurance, and other lines of commercial liability insurance, together with the exit of some insurers from those lines of business, has placed liability insurance on the public agenda. At the same time, asbestos and environmental losses continue to mount under general liability insurance policies sold long ago, when no one could have predicted the extent or cost of such losses. In combination, these and other related events have raised serious concerns about the insurability of liability risks and have prompted calls for dramatic efforts to roll back the advance of liability law in the U.S. and Europe. With this ‘‘crisis’’ as the background, this essay represents a preliminary effort to explore insurance for liability risks, with the goal of articulating a framework that will guide more systematic and empirical study. In the main part of this essay I describe a new conceptual framework for analysing liability insurance risks. The framework helps explain why predicting losses can be so difficult in the liability insurance context and why cycles in insurance pricing are more extreme in liability insurance than in other lines of insurance. Much of this difficulty is already understood to follow from the ‘‘long tail’’ or ‘‘duration of liability’’ problem in liability insurance. The primary contribution of the new framework is breaking apart what I will call ‘‘liability developments risk’’ for separate analysis, facilitating a better understanding of when and why the duration of liability matters. In concluding, I use a brief analysis of medical liability insurance to offer some observations on insurability and to sound the call for systematic empirical research aimed at improving how liability insurance institutions manage liability developments risk.