ABSTRACT The stellar mass in the halo of the Milky Way is notoriously difficult to determine, owing to the paucity of its stars in the solar neighbourhood. With tentative evidence from Gaia that the nearby stellar halo is dominated by a massive accretion event – referred to as Gaia-Enceladus or Sausage – these constraints are now increasingly urgent. We measure the mass in kinematically selected mono-abundance populations (MAPs) of the stellar halo between −3 < [Fe/H] < −1 and 0.0 < [Mg/Fe] < 0.4 using red giant star counts from APOGEE DR14. We find that MAPs are well fit by single power laws on triaxial ellipsoidal surfaces, and we show that that the power-law slope α changes such that high [Mg/Fe] populations have α ∼ 4, whereas low [Mg/Fe] MAPs are more extended with shallow slopes, α ∼ 2. We estimate the total stellar mass to be $M_{*,\mathrm{tot}} = 1.3^{+0.3}_{-0.2}\times 10^{9}\ \mathrm{M_{\odot}}$, of which we estimate ${∼}0.9^{+0.2}_{-0.1} \times 10^{9}\ \mathrm{M_{\odot}}$ to be accreted. We estimate that the mass of accreted stars with e > 0.7 is M*,accreted, e > 0.7 = 3 ± 1 (stat.) ± 1 (syst.) × 108 M⊙, or ${∼}30{-}50{{\ \rm per\ cent}}$ of the accreted halo mass. If the majority of these stars are the progeny of a massive accreted dwarf, this places an upper limit on its stellar mass, and implies a halo mass for the progenitor of ∼1010.2 ± 0.2 M⊙. This constraint not only shows that the Gaia-Enceladus/Sausage progenitor may not be as massive as originally suggested, but that the majority of the Milky Way stellar halo was accreted. These measurements are an important step towards fully reconstructing the assembly history of the Milky Way.