We use elastic block modelling of 126 GPS site velocities from Jamaica, Hispaniola, Puerto Rico and other islands in the northern Caribbean to test for the existence of a Hispaniola microplate and estimate angular velocities for the Gônave, Hispaniola, Puerto Rico-Virgin Islands and two smaller microplates relative to each other and the Caribbean and North America plates. A model in which the Gônave microplate spans the whole plate boundary between the Cayman spreading centre and Mona Passage west of Puerto Rico is rejected at a high confidence level. The data instead require an independently moving Hispaniola microplate between the Mona Passage and a likely diffuse boundary within or offshore from western Hispaniola. Our updated angular velocities predict 6.8 ± 1.0 mm yr−1 of left-lateral slip along the seismically hazardous Enriquillo-Plantain Garden fault zone of southwest Hispaniola, 9.8 ± 2.0 mm yr−1 of slip along the Septentrional fault of northern Hispaniola and ∼14–15 mm yr−1 of left-lateral slip along the Oriente fault south of Cuba. They also predict 5.7 ± 1 mm yr−1 of fault-normal motion in the vicinity of the Enriquillo-Plantain Garden fault zone, faster than previously estimated and possibly accommodated by folds and faults in the Enriquillo-Plantain Garden fault zone borderlands. Our new and a previous estimate of Gônave-Caribbean plate motion suggest that enough elastic strain accumulates to generate one to two Mw∼ 7 earthquakes per century along the Enriquillo-Plantain Garden and nearby faults of southwest Hispaniola. That the 2010 M= 7.0 Haiti earthquake ended a 240-yr-long period of seismic quiescence in this region raises concerns that it could mark the onset of a new earthquake sequence that will relieve elastic strain that has accumulated since the late 18th century.