The determination of the Kobayashi-Maskawa matrix element Vus from hyperon β decays has long had a hidden uncertainty due to the almost universal assumption of SU(3) invariance in Cabibbo-type fits, especially since the data definitely indicate the presence of SU(3) breaking. We have reanalyzed the hyperon-decay data using the pattern of symmetry breaking predicted by the quark model including the center-of-mass correction. We find that the SU(3)-broken picture is far superior to the assumption of perfect SU(3), and provides a good fit to experiment. The sensitivity of Vus to the breaking is not large and we find Vus=0.220±0.001±0.003 (the errors are experimental and theoretical, respectively), in agreement with the results from kaon decay by Leutwyler and Roos.