An extensive set of geodetic observations from the 170-km-long central creeping segment of the San Andreas fault is used to constrain the broadscale deformation pattern, determine the distribution of aseismic slip along the fault, and estimate the rate of relative plate motion across central California. Although small but significant departures appear to exist, rigid block displacements across the major faults provide a good first-order approximation to the deformation, as concluded previously by Savage and Buford. Geodetic measurements of ~60-km aperture that cross the fault zone provide a firm lower bound of 33 ¿ 1 mm/yr for the rate of right-lateral plate motion parallel to the San Andreas fault system since 1885. Measurements near the fault indicate rates of movement close to but still significantly less than this lower bound rate, ranging from 25¿1 mm/yr near San Juan Bautista-Hollister to 29¿mm/yr farther southeast on the fault. There is no evidence for changes in the grossmovement rate since 1885; in particular, no significant rate changes spanned the occurrence time of the great 1906 earthquake to the north, indicating that the creeping zone largely isolates the episodic movements occurring on the two adjacent locked sections of the fault. Further refinements in modeling the deformation are provided by making plausible assumptions on the fault geometry of the creep zone and the conditions appropriate for the locked portions of the fault, and using a simple dislocation model, all available data can be inverted simultaneously. Shallow slippage is close to the maximum values of observed surface creep but is smoother as a function of distance along the fault, and rates vary from 20 to 30 mm/yr. The deep slip rate, the relative plate motion rate for this model geometry, is uniquely determined in the inversion, but its precise value is sensitive to the model assumptions. However, off-fault data do constrain its value to be no greater than about 45 mm/yr. Although the data do not require it, modest amounts of elastic strain accumulation on the creeping segment of the fault cannot be unambiguously precluded. The difference between the near-fault and longer-range movement rates, or equivalently, the deficit between deep and shallow slip rates, is not definitively explained, although it may be due to slip on minor faults or an effect of the locked ends on the fault. These differences suggest that small amounts of deformation occur off the major faults, and although this straining is too faint or inhomogeneous to be resolved with available data, its presence may be reflected in the diffuse regional off-fault seismicity.