The original observation that reduced heart rate variability (HRV) confers poor prognosis after myocardial infarction has been followed by many studies of heart rate dynamics. We tested the hypothesis that an entropy-based local dynamics measure gave prognostic information in ambulatory patients undergoing 24-h electrocardiography. In this context, entropy is the probability that short templates will find matches in the time series. We studied RR interval time series from 24-h Holter monitors of 1564 consecutive patients over age 39. We generated histograms of the count of templates as a function of the number of templates matches in short RR interval time series, and found characteristic appearance of histograms for atrial fibrillation, sinus rhythm with normal HRV, and sinus rhythm with reduced HRV and premature ventricular contractions (PVCs). We developed statistical models to detect the abnormal dynamic phenotype of reduced HRV with PVCs and fashioned a local dynamics score (LDs) that, after controlling for age, added more prognostic information than other standard risk factors and common HRV metrics, including, to our surprise, the PVC count and the HRV of normal-to-normal intervals. Addition of the LDs to a predictive model using standard risk factors significantly increased the ROC area and the net reclassification improvement was 27%. We conclude that abnormal local dynamics of heart rate confer adverse prognosis in patients undergoing 24-h ambulatory electrocardiography.