We present new results from a paleoseismic trenching campaign at a site across the Jordan Gorge Fault (JGF), the primary strand of the Dead Sea Transform in northern Israel. In addition to the previously recognized earthquakes of 1202 and 1759 C.E., we observe evidence for eight surface-rupturing earthquakes prior to the second millennium C.E. The past millennium appears deficient in strain release with the occurrence of only two large ruptures, when compared with the preceding 1200 years. Assuming Gutenberg-Richter magnitude-frequency distribution, there is a discrepancy between measured rate of small-magnitude earthquakes (M <4) from instrumental records and large earthquake rates from paleoseismic records. The interevent time of surface-rupturing earthquakes varies by a factor of two to four during the past 2 ka at our site, and the fault's behavior is not time predictable. The JGF may be capable of rupturing in conjunction with both of its southern and northern neighboring segments, and there is tentative evidence that earthquakes nucleating in the Jordan Valley (e.g., the 749 C.E. earthquake) could either rupture through the stepover between the faults or trigger a smaller event on the JGF. We offer a model of earthquake production for this segment in which the long-term slip rate remains constant while differing earthquake sizes can occur, depending on the segment from which they originated and the time since the last large event. The rate of earthquake occurrence in this model does not produce a time-predictable pattern over a period of 2 ka as a result of the interplay between fault segments to the south and north of the JGF.