CPU overloads are generally a result of unexpected events or trying to do too much with too little. The default configuration of the xPC Target real-time kernel will stop model execution when a CPU overload is detected. This guarantees hard real-time performance. Nevertheless, it is possible to continue model execution in the presence of CPU overloads. Also, blocks are available that can sense and control the CPU overload counter. This allows for implementing exception handling when CPU overloads occur. These capabilities are described and test models are included to demonstrate.