Precise geodetic measurement of coseismic crustal movements is one of the main sources of information on faults that ruptured in earthquakes. Such information is provided routinely by a Global Positioning System (GPS) array, e.g. within a day in Japan. Interferometric Synthetic Aperture Radar (InSAR) is often used to reinforce the fault model, and newly launched Japanese Advanced Land Observing Satellite (ALOS), named “Daichi,” has enhanced such chances with its dedicated L-band radar. A GPS array has been an important tool to map secular crustal deformation, helping us identify local-scale strain accumulation heterogeneity (e.g. detection “strain concentration zone” in Central Japan) as well as large-scale tectonic plate movement. In this keynote presentation I review past and present contributions of precise geodetic measurements, positioning and gravimetry, to earthquake physics. One such contribution is the discovery and observations of slow earthquakes that do not radiate waves detectable with seimometers. Repeating slow slip events were found in several subduction zones at transient depths between seismogenic and stably sliding plate interfaces. General characterization of such events ranging from small non-volcanic deep tremors to large-scale afterslips has been done as a new category of faulting with a scaling law different from regular earthquakes. Current issues include the origin of apparent periodicity of such events, i.e. if external rhythms (e.g. seasonal loads and pole tides) control their recurrences, and I introduce a new example of repeating slow events in the Ryukyu Trench, where the fastest plate convergence in the world (12 cm/yr) takes place. Recent advent of gravity satellites enabled a gravimetric approach to earthquakes. For the 2004 Sumatra-Andaman Earthquake, coseismic geoid depression and its postseismic recovery have been detected for the first time by GRACE. This is the first opportunity to use two different (displacement and gravity change) approaches to study postseismic processes at depth, which will significantly reduce the non-uniqueness in inferring physical processes after large earthquakes.