Many physical processes have distributions of times between events that hav
e non-normalizable, power law probability density functions (PDF's). The mo
ments of such distributions are not defined. We found that the PDF's of the
times between events of ventricular tachyarrhythmia (rapid heart rate) and
premature ventricular contractions have a power law form indicative of a n
on-normalizable distribution, and that the timing between these events cann
ot be meaningfully characterized by the mean frequency of such events. The
Hurst analysis showed that there were self-similar correlations in the data
. These results indicate that the physical processes that disrupt the norma
l rhythm of the heart produce a fractal pattern in the timing between these
events. It also suggests that the mean and the variance of the frequency o
f these events may not be good measures to assess the status of patients wi
th these arrhythmias and determine the effectiveness of therapeutic procedu
res. [S1063-651X(98)01209-3].