We present weak-lensing mass measurements of 50 X-ray luminous galaxy clusters at 0.15 z 0.3, based on uniform high quality observations with Suprime-Cam mounted on the 8.2-m Subaru telescope. We pay close attention to possible systematic biases, aiming to control them at the ∼ < 4 per cent level. The dominant source of systematic bias in weak-lensing measurements of the mass of individual galaxy clusters is contamination of background galaxy catalogues by faint cluster and foreground galaxies. We extend our conservative method for selecting background galaxies with (V − i ′ ) colours redder than the red sequence of cluster members to use a colour-cut that depends on cluster-centric radius. This allows us to define background galaxy samples that suffer 1 per cent contamination, and comprise 13 galaxies per square arcminute. Thanks to the purity of our background galaxy catalogue, the largest systematic that we identify in our analysis is a shape measurement bias of 3 per cent, that we measure using simulations that probe weak shears upto g = 0.3. Our individual cluster mass and concentration measurements are in excellent agreement with predictions of the mass-concentration relation. Equally, our stacked shear profile is in excellent agreement with the Navarro Frenk and White profile. Our new LoCuSS mass measurements are consistent with the CCCP and CLASH surveys, and in tension with the Weighing the Giants at ∼ 1 − 2σ significance. Overall, the consensus at z 0.3 that is emerging from these complementary surveys represents important progress for cluster mass calibration, and augurs well for cluster cosmology.