Economic models of tort law evaluate the efficiency of liability rules in terms of care and activity levels. A liability regime is optimal when it creates incentives to maximize the value of risky activities net of accident and precaution costs. The allocation of primary and residual liability allows policy makers to induce parties to undertake socially desirable care and activity levels. Traditionally, tort law systems have assigned residual liability either entirely on the tortfeasor or entirely on the victim. In this paper, we unpack the cheapest-cost-avoider principle to consider the virtues and limits of loss-sharing rules in generating optimal (second-best) incentives and allocations of risk. We find that loss sharing may be optimal in the presence of countervailing policy objectives, homogeneous risk avoiders, and subadditive risk, which potentially offers a valuable tool for policy makers and courts in awarding damages in a large number of real-world accident cases.