Abstract Previous workers, mainly mapping on-land active faults on Caribbean islands, defined the northern Caribbean plate boundary zone as a 200-km-wide and bounded by two active and parallel strike-slip faults: the Oriente fault along the northern edge of the Cayman trough having a GPS rate of 14 mm/yr, and the Enriquillo-Plaintain Garden fault zone (EPGFZ) having a rate of 5-7 mm/yr. In this study, we used 5,000 km of industry and academic data from the Nicaraguan Rise south and southwest of the EPGFZ in the maritime areas of Jamaica, Honduras, and Colombia to define an offshore, 700-km-long, active, left-lateral strike-slip fault in what had been considered previously the stable interior of the Caribbean plate as determined from plate-wide GPS studies. The fault was named by previous workers as the Pedro Banks fault zone (PBFZ) because a 100-km-long segment of the fault forms an escarpment along the Pedro carbonate bank of the Nica-raguan Rise. Two fault segments of the PBFZ are defined: the 400-km-long eastern segment that exhibits large negative flower structures 10-50 km in width; fault segments rupture the sea floor as defined by high resolution 2D seismic data; and a 300-km-long western segment that is defined by a narrow zone of anomalous seismicity first observed by previous workers. The western end of the PBFZ terminates on a Quaternary rift structure, the San Andres rift, associated with Plio-Pleistocene volcanism and thickening trends indicating initial rifting in the late Miocene. The southern end of the San Andreas rift terminates on the western Hess fault which also exhibits active strands consistent with left-lateral, strike-slip faults. The total length of the PBFZ-San Andres rift-Southern Hess escarpment fault is 1,200 km and traverses the entire western end of the Caribbean plate. Our interpretation is similar to previous models that have proposed the “stable” western Caribbean plate is broken by this fault, the rate of displacement of which is less than the threshold recognizable from the current GPS network (~3 mm/yr). The late Miocene age of the fault indicates it may have activated during the late Miocene to recent His-paniola-Bahamas oblique collision event.