This paper illustrates an algorithm for real time detection QRS complex from ECG signal for computation of heart rate. The algorithm is implemented on a standalone embedded system based on Atmel 89C51 microcontroller. Synthetic ECG is generated using Physionet data through the parallel port (LPT1) of a personal computer and delivered to the embedded system. During an initial training period of first 1500 samples, some amplitude and slope based signatures are learned to form a rule base, which are used for detecting the subsequent QRS regions accurately. An average sensitivity of 97.82% and predictivity of 98.35% respectively are obtained from MIT BIH arrhythmia data. From the detected successive R peak locations heart rate has been computed. © 2012 IEEE.