We present mass models of the Milky Way created to fit observational constraints and to be consistent with expectations from theoretical modelling. The method used to create these models is that demonstrated in McMillan (2011), and we improve on those models by adding gas discs to the potential, considering the effects of allowing the inner slope of the halo density profile to vary, and including new observations of maser sources in the Milky Way amongst the new constraints. We provide a best fitting model, as well as estimates of the properties of the Milky Way. Under the assumptions in our main model, we find that the Sun is $R_0 = (8.20\pm0.09)\,\mathrm{kpc}$ from the Galactic Centre, with the circular speed at the Sun being $v_0 = (232.8\pm3.0)\,\mathrm{km}\,\mathrm{s}^{-1}$; that the Galaxy has a total stellar mass of $(54.3\pm5.7)\times10^9\,{\rm M}_\odot$, a total virial mass of $(1.30 \pm 0.30)\times10^{12}\,{\rm M}_\odot$ and a local dark-matter density of $0.38\pm0.04\,\mathrm{GeV\,cm}^{-3}$, where the quoted uncertainties are statistical. These values are sensitive to our choice of priors and constraints. We investigate systematic uncertainties, which in some cases may be larger. For example, if we weaken our prior on $R_0$, we find it to be $(7.97\pm0.15)\,\mathrm{kpc}$ and that $v_0=(226.8\pm4.2)\,\mathrm{km}\,\mathrm{s}^{-1}$. We find that most of these properties, including the local dark-matter density, are remarkably insensitive to the assumed power-law density slope at the centre of the dark-matter halo. We find that it is unlikely that the local standard of rest differs significantly from that found under assumptions of axisymmetry. We have made code to compute the force from our potential, and to integrate orbits within it, publicly available.