ABSTRACT We measure the total stellar halo luminosity using red giant branch (RGB) stars selected from Gaia data release 2. Using slices in magnitude, colour, and location on the sky, we decompose RGB stars belonging to the disc and halo by fitting two-dimensional Gaussians to the Galactic proper motion distributions. The number counts of RGB stars are converted to total stellar halo luminosity using a suite of isochrones weighted by age and metallicity, and by applying a volume correction based on the stellar halo density profile. Our method is tested and calibrated using Galaxia and N-body models. We find a total luminosity (out to 100 kpc) of $L_{\rm halo} = 7.9 ± 2.0 \times 10^8\, \mathrm{L}_⊙$ excluding Sgr, and $L_{\rm halo} = 9.4 ± 2.4 \times 10^8\, \mathrm{L}_⊙$ including Sgr. These values are appropriate for our adopted stellar halo density profile and metallicity distribution, but additional systematics related to these assumptions are quantified and discussed. Assuming a stellar mass-to-light ratio appropriate for a Kroupa initial mass function (M⋆/L = 1.5), we estimate a stellar halo mass of $M^⋆ _{\rm halo} = 1.4 ± 0.4\times 10^9 \, \mathrm{M}_⊙$. This mass is larger than previous estimates in the literature, but is in good agreement with the emerging picture that the (inner) stellar halo is dominated by one massive dwarf progenitor. Finally, we argue that the combination of a ${∼}10^9\, \mathrm{M}_⊙$ mass and an average metallicity of 〈[Fe/H]〉 ∼ −1.5 for the Galactic halo points to an ancient (∼10 Gyr) merger event.