Several analyses of SSH-SH differential travel-times disagree on whether these data contain evidence for azimuthal anisotropy in the North Atlantic upper mantle. To explore possible explanations for these discrepant results, we calculated the azimuthal variation in SSH-SH travel-times predicted by an anisotropic upper mantle model. Theoretical travel-times were obtained by cross-correlating S and SS phases on transverse-component reflectivity synthetic seismograms. We determined that predicted azimuthal travel-time patterns are sensitive to the relative amplitudes of radial and transverse particle motions incident on the anisotropic medium. If a large variation in the relative incident amplitudes exists in a dataset, the overlap of different azimuthal travel-time patterns may obscure anisotropic signatures. We analyzed shear-wave splitting in a dataset of 12 SS and 10 S phase observations to constrain the presence of anisotropy in the North Atlantic upper mantle. A comparison of SS splitting parameters with S and SKS particle motions suggests that shear wave splitting apparent in the SS phases is at least partially due to anisotropy in the vicinity of their bounce-points, although contributions from anisotropy in the source and station regions cannot be ruled out. Observed SS splitting is consistent with an olivine model in which horizontal a-axes are aligned at an average of N24¿W in the centrnal North Atlantic (15¿--40¿), and at an average of N23¿E in the northern North Atlantic (40¿--50¿N).