Addition of the spreading rates in the back-arc basins of the Lau Basin–Havre Trough and the Mariana Trough to the absolute motions of the Indian and Philippine plates, respectively, provides an estimate of the absolute motion of the Tonga-Kermadec and Mariana trenches. The trenches are moving toward the Pacific Basin for all absolute-motion models. Back-arc spreading is considered an unlikely cause for the trench motion, given the seismic evidence for poor coupling at the trenches. Thus, the cause of spreading at these arcs must rest at least partly with motion of the lower plate. Rollback due to negative buoyancy is rejected as a cause for this motion because it fails to explain the recent onset of the spreading. No-net-torque reference frames for 5 and 10 Ma indicate that the rate of subduction of the Pacific plate at each of these arcs may have increased by 2–3 cm/yr between 5 and 10 Ma, providing support for the model of Furlong et al., in which back-arc spreading is initiated by adjustment of the slab profile to a change in subduction rate.