Global positioning system (GPS) receivers are capable of generating precise geodetic data that can yield important constraints on both plate kinematics and dynamics. Geodetic measurements can determine plate motion rates over time scales for which little data are currently available. High-quality geodetic data may also allow the investigation of plate driving forces because changes in the intraplate stress field can potentially be inferred from measurement of the resulting crustal strain. A measurement program designed to investigate kinematic and dynamic aspects of plate tectonics should be concentrated in the Pacific region. This area contains the largest and fastest moving plate, up to 17 cm/year. Furthermore, subduction zones, which play a quantitatively important role in plate driving forces, are largely restricted to the Pacific region. We summarize accuracy studies showing that for short (< 100 km) baselines, centimeter-level accuracy can be expected using only mobile stations. For longer baselines, uncertainty in the orbit ephemeris of the GPS satellites is a major error source. Performing simultaneous observations at widely separated (~3000 km) fiducial stations near the region of interest, however, should allow centimeter-level accuracy for baselines up to several thousand kilometers in length. This performance level is predicated upon the assumption that fiducial baselines are known a priori to the centimeter level, for example, from very-long-baseline interferometry, and that corrections for tropospheric path delay are accurate to the centimeter level. The fiducial network location is flexible, limited mainly by the requirement for mutual satellite visibility.