Precise measurements of spreading rates on marine magnetic profiles collected to the west of the Antarctic Peninsula have enabled some consideration of the forces governing plate motion, since Antarctic—Phoenix motion has been controlled by the local rather than global force balance over the last 35 m.y. The total effective driving force per unit length of trench is calculated to have ranged between 2.6 and 3.6×1012 N/m, which is much less than is commonly thought necessary to support subduction. Conventional calculations may overestimate slab pull for old slabs because they neglect the effect of extensional disruption in limiting the contribution to the balance of forces at the trench. The low estimate of driving forces obtained here implies that resistive forces are also smaller than is generally assumed. Driving forces show a strong correlation with observed spreading rates, which indicates that resistive forces were largely velocity dependent. Fluid migration up the subduction zone may elevate temperatures in and around the shear zone, reducing resistive forces below the levels required by purely conductive models. Changes in convergence rate may affect the depths of both the brittle/ductile deformation boundary and the basalt/eclogite phase change, causing a negative feedback which would appear as a velocity‐dependent resistive force. The different driving forces acting on the NE and SW parts of the Phoenix plate, as a consequence of older oceanic lithosphere at the trench in the NE, caused Antarctic–Phoenix spreading to take place about a near pole to the SW since 21 m.y. ago, and ultimately resulted in disruption of the Phoenix plate about 9 m.y. ago. Spreading rates decreased abruptly about 6 m.y. ago, probably because of E‐W compression across the long transform faults bounding the Phoenix plate. However, spreading on the last three segments of the Antarctic–Phoenix Ridge continued at least until 4 m.y. ago. Either spreading stopped progressively from SW to NE, or the final stage took place about a very near pole to the SW. A magnetic quiet zone extends up to 95 km from the margin between the Tula Fracture Zone and the North Anvers Fracture Zone, and is thought to indicate that the ridge crest became buried by terrigenous sediment prior to collision. The absence of a magnetic quiet zone associated with the most recent ridge crest–trench collisions suggests a change in sedimentary regime during the late Miocene. Anomalously fast apparent spreading rates between 23 and 21 m.y. ago are thought to indicate an error in this part of the magnetic reversal time scale.