The recent financial turmoil highlights the incentive of highly leveraged financial institutions to take excessive risk, given the protection of limited liability. During the nineteenth and early twentieth century, many banks operated under liability rules which obligated shareholders to bear larger costs of bank insolvency in the form of contingent, or even unlimited, liability. This article examines the empirical relationship between the size of banks' contingent liability and their risk-taking behaviour using data on British banks from 1878 to 1912. We find that banks with more contingent liability appear to have taken less risk. We also find evidence that the risk-reducing effects of contingent liability were larger for banks with higher leverage, suggesting that contingent capital mitigated the moral hazard problem at banks.