Strange-quark mass from combined e + e − and τ -decay data: test of the isospin symmetry and implications on ǫ ′ /ǫ and m u,d Abstract I extract the strange-quark mass using a τ -like decay sum rule for the φ-meson, and some other sum rules involving its difference with the vector component of the hadronic τ -decay. As a conservative estimate, one obtains to order α 3 s : m s (1 GeV) = (178 ± 33) MeV =⇒ m s (2 GeV) = (129 ± 24) MeV, while the positivity of the spectral function leads to the upper bound: m s (1 GeV) ≤ (200 ± 28) MeV =⇒ m s (2 GeV) ≤ (145 ± 20) MeV. These results are in good agreement with the existing sum rule and τ -decay results, and, in particular, with the result from the the sum rule involving the difference of the isoscalar and isovector components of the e + e − → hadrons data. This signals small effects of the SU (2) isospin violation due to the ω-ρ mixing parameters, and questions the reliability of the existing sum rule estimates of these parameters. Combining our result with the recent data on ǫ ′ /ǫ, we can estimate, within the standard model, the four-quark weak matrix elements B 1/2 6 − 0.54B 3/2 8 to be about (2.8 ± 1.3). This result may suggest a large violation of the vacuum saturation estimate similarly to the case of the four-quark condensates obtained from the sum rules analysis, and can serve as a guide for a future accurate non-perturbative extraction of such matrix elements. Combining our result with the sum rule estimate of m u + m d and with the Dashen formula for the mass ratio, one can deduce the update values: m d (2 GeV) = (6.4 ± 1.1) MeV and m u (2 GeV) = (2.3 ± 0.4) MeV.