On the basis of the net transfer reactions among garnet, biotite, plagioclase and quartz (for both Mg and Fe end-member models), the garnet–biotite–plagioclase–quartz (GBPQ) geobarometer was empirically calibrated under physical conditions of P = 1·0–11·4 kbar and T = 515–878°C, based on the input garnet–biotite temperatures and garnet–aluminosilicate–plagioclase–quartz (GASP) pressures of 224 natural aluminosilicate-bearing metapelitic samples collated from the literature. The calibrations are internally consistent with the asymmetric quaternary solid solution model of garnet, the symmetric quaternary solid solution model of biotite, and the Al-avoidance ternary solid solution model of plagioclase in calibrating the garnet–biotite geothermometer and the GASP geobarometer. The resulting two GBPQ barometer formulae reproduce the input GASP pressures well within ±1·0 kbar (mostly within ±0·5 kbar). For both aluminosilicate-bearing and aluminosilicate-absent metapelites, the two GBPQ barometry formulae yielded identical pressures, whether the sample was included or not included in calibrating the GBPQ barometry. The random error of the GBPQ barometry may be expected as ±1·2 kbar. The dP/dT slopes of these two GBPQ formulae are close to that of the GASP barometer in P–T space. Applications of the GBPQ barometry of aluminosilicate-absent metapelites to the rocks within a thermal contact aureole, or rocks within a limited geographical area without post-metamorphic structural discontinuity, show no obvious pressure change. It may be concluded that the two GBPQ barometry formulae derived in this study may be used as practical tools for metamorphic pelites under the conditions of 515–878°C and 1·0–11·4 kbar, in the composition range of X_gros >3% in garnet, X_an >17% in plagioclase, and in biotite. Application of the GBPQ barometer beyond the calibration range should be undertaken with caution.