Deposited in DRO : 19 February 2019 Version of attached le : Peer-review status of attached

We present and apply a method to infer the mass of the Milky Way (MW) by comparing the dynamics of MW satellites to those of model satellites in the EAGLE cosmological hydrodynamics simulations. A distribution function (DF) for galactic satellites is constructed from EAGLE using specific angular momentum and specific energy, which are scaled so as to be independent of host halo mass. In this two-dimensional space, the orbital properties of satellite galaxies vary according to the host halo mass. The halo mass can be inferred by calculating the likelihood that the observed satellite population is drawn from this DF. Our method is robustly calibrated on mock EAGLE systems. We validate it by applying it to the completely independent suite of 30 AURIGA high-resolution simulations of MW-like galaxies: the method accurately recovers their true mass and associated uncertainties. We then apply it to 10 classical satellites of the MW with six-dimensional phase-space measurements, including updated proper motions from the Gaia satellite. The mass of the MW is estimated to be MMW 200 = 1.17+0.21 −0.15 × 1012 M (68 per cent confidence limits). We combine our total mass estimate with recent mass estimates in the inner regions of the Galaxy to infer an inner dark matter (DM) mass fraction MDM(< 20 kpc)/MDM 200 = 0.12, which is typical of ∼1012 M lambda cold dark matter haloes in hydrodynamical galaxy formation simulations. Assuming a Navarro, Frenk and White (NFW) profile, this is equivalent to a halo concentration of cMW 200 = 10.9+2.6 −2.0.