ECG monitoring devices for coronary care units require reliable R-wave dectors as a prerequisite for subsequent analysis. A simple algorithm has been developed for the detection of R-waves in an ECG signal with a microprocessor (Motorola MC6800). Detection criteria used were gradient and duration of a single upslope or downslope. Rejection of muscle noise and other artifiacts was achieved by setting system constants to optimum values which were determined by a short preliminary trial. This constant threshold system has been shown to cope with sudden changes of QRS complex amplitude better than some self adjusting systems. Assessment runs using patient tapes from a coronary care unit show total errors due to false positives and negatives are typically about 1% of the actual number of R-wave occurrences.