Ventricular ectopic beats are clinically important because of their association with the problem of sudden cardiac death. Ventricular ectopic activity (VEA) is most commonly evaluated using long-term ambulatory electrocardiographic (ECG) recordings. The results of these recordings are usually summarized in the form of hourly counts of VEA events.Although such hourly counts are useful, they are a nonspecific measure which, in effect, treats VEA generation as an unpredictable process.This thesis develops a new means for characterizing the activity of the VEA-generating mechanism using the beat type and timing information of the long-term ECG. The approach is founded on the hypothesis that the VEA mechanism is often stable over time, although its activity may be modulated by physiologic influences and by the timing of the sino-atrial node impulses.This hypothesis implies that there should be some relationship between the beat type and timing of preceding beats and the subsequent generation of ectopic beats.The VEA generation was characterized by a multi-dimensional conditional distribution which gives the probability that an ectopic beat will follow a specific conditioning sequence of beat types and intervals. Using this approach a three part study was performed. The first step examined the conditional distributions and tested the null hypothesis that the generation of ectopic beats is unpredictable.Using a database of half-hour ECG tapes from 66 patients, the null hypothesis was rejected at p=0.005 in 80% of the tapes.The second step developed a means of displaying the multidimensional distributions. By an appropriate transformation of the conditioning variables, the distributions could be represented as one-dimensional distributions. The resulting distributions were useful for. separating patients into groups and for finding characteristic, VEA-related events in the ECG data.The third step examined the reproducibility of the onedimensional distributions over time. Using ten-hour ECG tapes, it was found that in 7 of 10 patients the patient-specific distribution, although modulated in amplitude, maintained a similar shape over the ten hours.The VEA characterization which has been developed goes significantly beyond the simple counting of ectopic beats. The distributions can be considered a kind of "fingerprint" of the VEA-generating mechanism -a "fingerprint" which may be useful in the selection and evaluation of therapy.