The petrogenesis of the 2800 km3 of magma erupted as the youngest Toba Tuff has been investigated using experimental petrology and 40Ar/39Ar dating of biotite, sanidine, hornblende, and plagioclase from the tuff. We find that hornblende does not crystallize experimentally from the magma at temperatures and pressures indicated by the natural mineral assemblage. Hornblende is also not in isotopic equilibrium with biotite and sanidine, both of which grew experimentally. Hornblende thus appears xenocrystic, despite being a major phase in the tuff. Some plagioclase is also xenocrystic, on the basis of Ar isotopes, but others are probably phenocrystic, because plagioclase grew experimentally. Crystal clots of hornblende + plagioclase observed in the tuff suggest that the xenocrysts came from a common source, which was at least 1.5 Ma (the oldest hornblende 40Ar/39Ar age). Our results suggest that the Toba Tuff magma resided at nearly water-saturated pressures of 100–150 MPa and that xenocrysts were entrained as recently as 10 yr before the eruption. The ubiquitous presence of hornblende in the tuff indicates that entrainment occurred throughout the 2800 km3 of magma.