Abstract The issue of QRS complex detection and delineation has been severely tracked throughout the last three decades. Specifically, the approaches for QRS detection, the generations of these algorithms, and newly developing methods, reflect the evolution of the processing power of computers. With the evolution of the faster processing computers, researchers stopped worrying about computational costs and started focusing on detection and delineation accuracy. Despite the fact that the families of algorithms can be grouped according to the overall results obtained over standard databases, and the most cited and robust algorithms perform at a very high level, example, over 99% in sensitivity and positive predictivity, most only used MIT–BIH Arrhythmia Database or AHA Database. Actually, for a detailed comparison, a standard and annotated database needs to be used, so that we can compare automatic detections with manual annotations. However, considering signals extracted in several circumstances, we can identify strengths and weaknesses from different approaches. Therefore, in this chapter, we propose to compare the performance of different already published QRS detection algorithms under different scenarios: resting ECG (clean clinical signals), patient being immersed in his/her daily routine (as in the case of holters and tele-health devices), multichannel analysis (2 or more leads), single channel analysis, QRS enlargement, QRS morphological variation, and occurrence of intense variability in the intervals between beats and/or amplitudes of R waves, ventricular, and supraventricular arrhythmias. Our goal is to enhance and detail advantages and disadvantages of the most robust techniques and also propose computational solutions for specific problems/challenges.