Plate rigidity is a key assumption of the plate tectonics theory. The assumption allows us to study plate motion on Earth with great simplicity. However, recent developments in understanding plate cooling have put the assumption of plate rigidity to test. Kumar and Gordon [2009] advance the theory of horizontal thermal contraction of oceanic lithosphere as the ultimate test of plate rigidity and predict that relative intraplate velocities due to horizontal thermal contraction within the oceanic lithosphere of a tectonic plate can be as high as 3 – 10 mm a-1. Contractional strains have been predicted to vary inversely with age and hence most of the strains related with horizontal thermal contraction should be accumulated and dissipated by young oceanic lithosphere. Transform faults provide the necessary free boundaries across which most of the contractional stresses can be dissipated by ridge-parallel contraction of the oceanic lithosphere. The strike of the transform faults, earlier predicted to be parallel to the relative plate motion direction, thus, should be biased as a result of transform fault perpendicular contraction of oceanic lithosphere. I first improve the global transform fault dataset used in MORVEL [DeMets et al., 2010] and then calculate residuals between the observed transform fault azimuths and those predicted for the rigid oceanic plates. I find that on an average, for the six plate pairs with both left-lateral and right-lateral slipping transform faults, azimuths are biased by about 0.75°±0.36° clockwise for left-lateral slipping and by −0.73°±0.22° (= ±1 standard error) counter-clockwise for the right-lateral slipping. I, then investigate if this observed bias can be caused by horizontal thermal contraction. For the six selected pairs with both right-lateral and left-lateral slipping transform faults, the mean bias is predicted to be only 0.38° for only right-lateral slipping faults and −0.46° for only left-slipping transform faults. Thus the magnitudes of bias predicted by horizontal thermal contraction are ≈60% as large as the observed residual between the observed strikes of transform faults and the strikes observed for the assumption of plate rigidity. Thus we cannot exclude some normal faulting in transform fault valleys. A global analysis of plate motion based on the transform fault azimuths corrected for the predicted bias due to horizontal thermal contraction shows that the hypothesis of no horizontal thermal contraction can be rejected with at least 40% of the contractional strains causing the rotation of strikes of transform faults and the remaining being accommodated by normal faulting in transform fault valleys. Thus we conclude that plates are not rigid. Based on the predictions of horizontal thermal contraction, I further test its effect on intraplate velocities and build a 2D kinematic model of the Pacific lithosphere between the Rivera and the Heezen fracture zones to calculate the predicted intraplate relative velocity field due to horizontal thermal contraction. For the kinematic assumption that balances net contraction and extension across the fracture zones, young oceanic lithosphere along the Rivera fracture zone is predicted to have a contractional velocity of 2.6 mm a-1 in the South-Southeast direction. This is big enough to account for the misfit of 5±3 mm a-1 in the PA-NA-NB-AN plate circuit and not big enough to explain the misfit of 14±5 mm a-1 for the CO-PA-NZ plate circuit.