The detection and description of the evoked cardiac response (ECR) has received much attention in recent literature, with an increasing concentration upon the 'noise' problem of sinus arrhythmia (SA). A departure from extrapolated pre-stimulus activity, obtained by employing procedures of various complexities, has been proposed as an SA-free measure of the ECR. The use of such procedures without consideration of respiratory changes appears questionable. Simulation of HR protocols across a change in respiratory period indicates that spurious ECRs, similar in form to those in the published literature, may be produced by such techniques. A simple method of correcting the problem is outlined. © 1979.