GPS measurements in the western Shumagin segment of the Alaska subduction zone show no significant strain. The elastic dislocation model that best fits the data has the entire plate interface slipping freely. The simplest explanation is that steady slip at the plate convergence rate is occurring on the entire plate interface, although we cannot rule out the existence of a shallow locked patch close to the trench large enough to generate a M~7 earthquake. The 99% confidence upper bound on the down-dip width of the locked patch is 55 km assuming the trench is the the updip limit of the locked zone. Unless there is significant temporal variability in the locking of the plate interface or viscoelastic effects from a past earthquake, the Shumagin segment cannot by itself release enough moment to generate a great (M~8) earthquake. ¿ 1999 American Geophysical Union