Campaign GPS measurements during 2001-2003 in the transition between subduction and collision of the Banda arc reveal how strain is partitioned away from the trench and distributed to other parts of the arc-trench system. Genrich, et. al. (1996) conducted a GPS campaign (1992-1994) throughout the Eastern Sunda and Banda arcs that demonstrated partial accretion of the arc to the Australian plate. We reoccupied many of the sites from this earlier study and 7 additional stations, 3 of which are new benchmarks. Our study shortened many baselines and extended the observation epoch to ten years for many key stations. The resulting GPS velocity field for the active Banda arc-continent collision reveals: 1) several mostly rigid crustal blocks exist in the transition from subduction to collision, 2) relative to an Asian reference frame, most of these blocks move in the same direction as the Australian lower plate, but at different rates, 3) block boundaries may exist between the islands of Lombok and Komodo, Flores and Sumba, Savu and West Timor, and between Timor and Darwin, 4) the Timor Trough may account for at least 20 mm/yr of motion between Timor and Darwin, 5) a major transverse fault off the coast of West Timor separates the Savu/Flores/Sumba block from the Timor/Wetar Block and may account for variations in movement in Rote, 6) the Flores thrust moves the eastern Sunda arc north relative to Asia by decreasing amounts to the west, 7) the back-arc Wetar Thrust system takes up the majority of plate convergence between Australia and Asia, and 8) fault boundaries are not found between many blocks, such as various islands of the Sunda arc and forearc with different amounts of motion.