The postseismic relaxation (postseismic displacement less displacement that would have occurred at the preseismic rate) measured by GPS and leveling following the 1989 M_S7.1 Loma Prieta earthquake is reexamined. The temporal dependence of the relaxation over the first 1200 days postseismic is well described by 1 − e^−t/τ, where τ = 414 ± 92 days. (Quoted uncertainties are standard deviations.) That temporal dependence appears to be a linear function of the cumulative number of M > 2.5 aftershocks that have occurred. The relaxation is attributed to afterslip (1.56 ± 0.20 m dextral strike slip and 0.60 ± 0.04 m reverse slip) on the downdip extension (depth, 16–21 km) of the coseismic rupture plus a collapse (0.11 ± 0.02 m fault-normal displacement) of the rupture zone (depth, 5–16 km). Because the postseismic uplift was determined by leveling over a route with relief in excess of 1000 m, an allowance (18 ± 2 ppm of height above the base elevation) for excess (beyond corrections already applied) height-dependent error in the measured uplift was estimated simultaneously with the afterslip and collapse parameters. This new solution for afterslip and collapse on the plane of the rupture provides an alternative explanation to the suggestion by Bürgmann et al. (1997) that the postseismic deformation was due to afterslip on the coseismic rupture and on a nearby, shallow, 30° dipping thrust.