How well do we know the Kobayashi-Maskawa matrix?

We evaluate the Kobayashi-Maskawa (KM) matrix using a maximum-likelihood method which takes into account experimental errors on input data and "systematic errors" associated with theoretical approximations. We carefully reconsider the constraints from meson decays and mixings as well as $\mathrm{CP}$ violation. We derive values for all matrix elements with errors smaller than 0.008, show that the ratio $\left|\frac{U_{\mathrm{ub}}}{U_{\mathrm{cb}}}\right|$ must be bigger than 0.036, and that the phase angle $\delta$ of the KM representation is smaller than $\pi$. Surprisingly our conclusions depend very weakly on the value of the top-quark mass. Any experiment violating the constraints of our fitted KM matrix would imply physics beyond the standard model with three generations and a minimal Higgs sector.