Abstract We explore an overlooked phenomenon in mortgage markets: repayment of underwater mortgages. Using a sample of mortgages terminated between 2007 and 2016, we show that such repayment indeed occurs, and that it is affected by the same factors commonly used in studies of default: the magnitude of home equity and the borrower’s credit score, which captures default cost as well as liquidity. A novel insight is that underwater repayers, unlike most defaulters, are not liquidity constrained, providing a much cleaner environment to study default costs. We estimate lower bounds on these costs. Our results indicate that default costs are substantial.