Numerous observations pertaining to the magnitude 9.0 2011 Tohoku-oki earthquake (offshore Japan) have led to new understanding of subduction zone earthquakes. By synthesizing published research results and our own findings, we explore what has been learned about fault behavior and Earth rheology from the observation and modeling of crustal deformation before, during, and after the earthquake. Before the earthquake, megathrust locking models based on land-based geodetic observations correctly outlined the along-strike location of the future rupture zone. Their incorrect definition of the locking pattern in the dip direction demonstrates the need to model the effects of interseismic viscoelastic stress relaxation and stress shadowing. The observation of decade-long accelerated slip downdip of the future rupture zone raises new questions on fault mechanics. During the earthquake, seafloor geodetic measurements revealed huge coseismic displacements (up to 31 m). Modeling of bathymetry difference before and after the earthquake suggests >60 m of coseismic slip of the most seaward 40 km of the fault in the main rupture area, with the slip peaking at the trench. Large differences in shallow slip between published rupture models are due mainly to the near absence of near-trench deformation measurements, but model simplifications in fault and seafloor geometry also bear large responsibility. After the earthquake, seafloor geodetic measurements provided unambiguous evidence for the dominance of viscoelastic relaxation in short-term postseismic deformation. There is little deep afterslip in the fault area where the decade-long pre-earthquake slip acceleration is observed. Investigating the physical processes responsible for the complementary spatial distribution of pre-slip and afterslip calls for new scientific research.