Urea selective catalytic reduction (SCR) systems are a promising technology for helping to lower NOx emissions from diesel engines. These systems also require on-board diagnostic (OBD) systems to detect malfunctioning catalysts. Conventional OBD methodology for a SCR catalyst involves the measurement of NOx concentration downstream of the catalyst. However, considering future OBD regulations, erroneous diagnostics may occur due to variations in the actual environment. Therefore, to enhance OBD accuracy, a new methodology was examined that utilizes NH3 slip as a new diagnostic parameter in addition to NOx. NH3 slip increases as the NOx reduction performance degrades, because both phenomena are based on deterioration in the capability of the SCR catalyst to adsorb NH3. Furthermore, NH3 can be measured by existing NOx sensors because NH3 is oxidized to NO internally. To make use of NH3 slip, an estimation model was developed. NH3 reactions of desorption, consumption by NOx reduction, and oxidation were formulated, and the coefficients of these reactions were fitted to multiple regression models. NH3 slip is estimated as a function of NH3 loading and temperature. NH3 loading is calculated from the increase-decrease amount of NH3 due to these reactions. On-board models of normal and malfunctioning catalysts are operated in parallel, and the OBD algorithm chooses the appropriate diagnostic timing automatically when the NOx sensor output of a malfunctioning catalyst is expected to be larger than that of a normal catalyst. As a result, this model-based SCR OBD control system has higher detectability than the conventional system.