This paper analyzes the basic credit risk parameters in residential mortgage lending and its evaluation with focusing on loss given default. We develop the method for the loss given mortgage default evaluation based on the econometric model for the probability of mortgage default, approximation value of collateral and residual loan amount in analyzing time horizon that is not used previously. Empirical distribution for mortgage loss given default is constructed based on developed approach and information on government-issued residential mortgage loans in Russia during 2008–2012 years and it is asymmetric and bimodal. The ratio of loss given mortgage default to the expected interest income was analyzed for certain pools of mortgage loans such as loans with verified and unverified borrower’s income, with different loan-to-value ratio and loans of primary lenders and regional operator of Agency of Home Mortgage Lending. We found that loans with high share of borrowed assets had high loss given default, exposure at default and creditor expected interest income. We show the usefulness of mortgage insurance for such loans as the way of expected loss compensation and found empirical evidence for that. Credit risk at the portfolio level was calculated and it increased with increase in the total cost related to a judicial settlement of outstanding mortgage debt. It can be used for calculation of loan loss provisions.