Following the financial crisis, the share of non-performing loans has significantly increased, while the regulatory guidelines on the Internal-Ratings Based (IRB) approach for capital adequacy calculation related to defaulted exposures remains too general. As a result, the high-risk nature of these portfolios is clearly in danger of being managed in a heterogeneous and inappropriate manner by those financial institutions permitted to use the IRB system, with the consequent undue variability of Risk-Weighted Assets (RWA). This paper presents a proposal to construct Advanced IRB models for defaulted exposures, in line with current regulations, that preserve the risk sensitivity of capital requirements. To do so, both parameters Expected Loss Best Estimate (ELBE) and Loss Given Default (LGD) in-default are obtained, backed by an innovative indicator (Mixed Adjustment Indicator) that is introduced to ensure an appropriate estimation of expected and unexpected losses. The methodology presented has low complexity and is easily applied to the databases commonly used at these institutions, as illustrated by two examples. JEL Classification: C51, G21, G28, G32