Glacial isostatic adjustment (GIA) is the main cause of deformation in intraplate North America. Here we use up to 3,271 Global Positioning System station velocities to image this 3-D deformation across the entire plate. We apply a new robust strain rate estimation algorithm (median estimation of local deformation), which does not require the assumption that part of the plate is unaffected by GIA, an assumption we show to be false. Our results show extension in the area underneath the Laurentide ice sheet, contrasted with a semiannular belt of horizontal contraction of up to ~4 × 10
9 yr
1 around the former ice sheet. This contractional belt is kinematically linked to an ~1–2 mm yr
1 far-field horizontal motion directed toward the ice sheet. Our results, together with a new robustly imaged vertical velocity field, are consistent with GIA as the main cause of deformation, although the contractional strain rates around the former ice sheet and the far-field horizontal velocities are significantly higher than those predicted by the ICE6G_C(VM5a) model. This finding suggests that our results, including the location of reversal in sign of horizontal motion relative to the ice sheet, will be useful to reevaluate the mantle viscosity structure used by GIA models. Besides the GIA-attributed deformation, we find almost no other region with significant strain accumulation. A plate-scale spatial correlation between strain rate and seismicity is absent, suggesting that GIA is a limited driver of contemporary seismogenesis and that intraplate seismicity must be attributable to factors other than secular strain accumulation. Plain Language Summary The theory of plate tectonics says that tectonic plates move as rigid blocks along the Earth’s surface and that the Earth’s crust should only deform at the boundary between plates. However, the recent explosion in the number of high-precision Global Positioning System stations allowed us to capture some subtle deformation patterns inside the North American plate that only became apparent by a very careful analysis of the relative motions between thousands of stations. We found that most of the plate is moving at 1–2 mm/year towards central Canada. Consequently, around most of Canada there is a zone where the crust is contracting. Within Canada, the crust is extending outward and is moving upward rapidly. These patterns can be explained by the process of the crust and mantle still rebounding from a time when it was covered by a thick ice sheet about 16,000 years ago. The fact that this causes the land to move towards the former ice sheet is an unexpected result that will be useful in understanding the relaxation properties of the underlying mantle. Moreover, we found that earthquakes inside the North American plate do not occur where we see the crust deform, which leaves these events still enigmatic.