We use observed line length change rates and shallow fault slip rates at Parkfield, California to estimate the slip rate at depth on the San Andreas fault. One model featuring a nearly monotonic transition from creeping behavior on the northwest to locked behavior on the southeast fits the data acceptably well. An isolated asperity fits the data better, but not significantly. Evidence for the asperity comes from two lines connecting stations where mildly inconsistent data have been collected. Thus evidence for the asperity is marginal. Stored elastic energy southeast of Parkfield could produce an earthquake of magnitude up to 8.0 that would have displacement and rupture length similar to historic earthquakes of the same size. Smaller events are of course more likely, but the geodetic data do not favor a ''characteristic'' M=6 event. ¿ American Geophysical Union 1988