Surface-wave data previously used by the authors to constrain upper mantle anisotropy in the Pacific Ocean are re-examined after an error was found in the calculations of the Love-wave phase-velocities. Sub-lithospheric anisotropy is still required by the corrected data but it extends to shallower depth than we originally proposed. Using different parameterizations of the upper-mantle, we present strong evidence that the bottom of the anisotropic layer has been reached around 250 km in the Pacific. Beneath North-America, no revision of our data is required and anisotropy could extend to around 400 km-depth. Our results are in agreement with an upper mantle anisotropy due to preferred orientation of olivine crystals in an overall horizontal shear strain parallel to the present-day plane motion. We suggest that this preferred orientation of the crystals mainly occurs within the upper mantle low-velocity layer. ¿ American Geophysical Union 1988