Near-field observations of high-precision borehole strain and pore pressure, show no indication of coherent accelerating strain or pore pressure during the weeks to seconds before the 28 September 2004 M 6.0 Parkfield earthquake. Minor changes in strain rate did occur at a few sites during the last 24 hr before the earthquake but these changes are neither significant nor have the form expected for strain during slip coalescence initiating fault failure. Seconds before the event, strain is stable at the 10−11 level. Final prerupture nucleation slip in the hypocentral region is constrained to have a moment less than 2 × 1012 N m (M 2.2) and a source size less than 30 m. Ground displacement data indicate similar constraints. Localized rupture nucleation and runaway precludes useful prediction of damaging earthquakes. Coseismic dynamic strains of about 10 microstrain peak-to-peak were superimposed on volumetric strain offsets of about 0.5 microstrain to the northwest of the epicenter and about 0.2 microstrain to the southeast of the epicenter, consistent with right lateral slip. Observed strain and Global Positioning System (gps) offsets can be simply fit with 20 cm of slip between 4 and 10 km on a 20-km segment of the fault north of Gold Hill (M0 = 7 × 1017 N m). Variable slip inversion models using gps data and seismic data indicate similar moments. Observed postseismic strain is 60% to 300% of the coseismic strain, indicating incomplete release of accumulated strain. No measurable change in fault zone compliance preceding or following the earthquake is indicated by stable earth tidal response. No indications of strain change accompany nonvolcanic tremor events reported prior to and following the earthquake.