Designing real-time systems such as electrocardiogram (ECG) analyzers and evaluating their performance precisely is not an easy task. One of the difficulties is due to the general loss of performance that is frequently observed when implementing data processing algorithms on such systems and running the devices in new environments. Performance values are traditionally estimated theoretically or numerically (personal computer simulations) or are evaluated on real data but in a simulated environment. While the best performance measurements are accredited due to experiments driven in a real environment, the data at stake in this case are not entirely controlled. We propose in this paper the new concept of a virtual patient for real-time ECG analyzers. This concept enables a real environment to be created virtually by generating signals from a database composed of real acquired data. The analog signals generated will be seen by the analyzer as if they were coming from instantaneous electrical heart activities. This procedure provides a more correct assessment of the ECG analyzer performance than was previously possible. The test bench is useful for Holter manufacturers since it can deliver genuine performance values of the device.