In a recent paper (hep-ph/0511160), members of the HPQCD collaboration have presented the most precise determination of the light (up, down and strange) quark masses to date.
This required both extensive unquenched simulations of QCD using some of the lightest (and hence hardest to work with) quark masses used so far, and a massive perturbative calculation at the two-loop order. The perturbative calculation is needed in order to connect the lattice-regularized bare quark masses to the masses as defined in the usually quoted MSbar scheme. The bare-quark masses required as input to the perturbative calculation come from simulations performed by the MILC collaboration, who use a highly-efficient formalism with so-called ``staggered'' quarks, with three flavors of light quarks in the Dirac sea.
Putting all these ingredients together, they find the MSbar masses at a scale of 2 GeV to be $$m_s = 87(0)(4)(4)(0)$$ MeV, $$m_u = 1.9(0)(1)(1)(2)$$ MeV and $$m_d = 4.4(0)(2)(2)(2)$$ MeV. The respective uncertainties are from statistics, simulation systematics, perturbation theory, and electromagnetic/isospin effects.
This means that the errors on the still rather contentious strange quark mass, for which a number of incompatible results exist, have been greatly reduced. This is a very major result, and a great success for Lattice QCD.