The Stewart pegmatite-aplite dike in Pala, California (USA) is well known as a source of lithium, gem minerals, and unusual phosphate minerals. We reinterpret the petrogenesis of the dike based on a combination of new regional and detailed geochemical isotopic and textural data. The Stewart dike, like other pegmatites in the Pala district and other major pegmatite districts in the northern Peninsular Ranges batholith, is enclosed within gabbro/mafic tonalite. The 40Ar/39Ar method of dating on muscovite from the dike, and U/Pb dating of zircon from the gabbro yield essentially the same age. Initial 87Sr/86Sr is similar for the dike, 0.7042, and the gabbro, 0.7036–0.7037, indicating a juvenile and likely common source for both. The extreme mineralogic, lithologic, and textural variations within the dike are interpreted to have resulted from in situ mineral segregation, autometasomatism, and migration of volatiles within an essentially closed system. Contacts between the pegmatite dike and the host gabbro are diffuse. All previous interpretations of the Stewart pegmatite dike invoked an allogenic origin, formed by fluids derived externally from a nearby or distant granitic body, with the fluids subsequently migrating to, and intruding, their gabbro/mafic tonalite host. These previously proposed sources of pegmatite-forming fluid(s) are ruled out here on the basis of similar ages of the host and pegmatite, similar initial 87Sr/86Sr isotopes, unrealistic 25–40 km distances between any proposed potential source and the host gabbro-tonalite pluton, and mineralogic and textural relationships indicating internal differentiation. For these reasons, we suggest an autogenic origin, whereby the gem-bearing pegmatites are petrogenetically limited to their gabbroic hosts and likely represent late-stage, internal segregation of magmatic fluids. We note that nearly every gem-bearing pegmatite in the northern Peninsular Ranges batholith is hosted within gabbro/mafic tonalite bodies, not within the more silicic plutons.