It is well established that the cratonic subcontinental lithospheric mantle (C-SCLM) represents a residue of extensively melted peridotite. The widespread occurrence of garnet in C-SCLM remains a paradox because experiments show that it should be exhausted beyond ∼20% melting. It has been suggested that garnet may have formed by exsolution from Al-rich orthopyroxene; however, the few documented examples of garnet exsolution in cratonic samples are exotic and do not afford a direct link to garnet in granular harzburgite. We report crystallographic, petrographic, and chemical data for an exceptionally well preserved orthopyroxene megacryst containing garnet lamellae, juxtaposed against granular harzburgite. Garnet lamellae are homogeneously distributed within the host orthopyroxene and occur at an orientation that is unrelated to orthopyroxene cleavage, strongly indicating that they formed by exsolution. Garnet lamellae are subcalcic Cr-pyrope, and the orthopyroxene host is high-Mg enstatite; these phases equilibrated at 4.4 GPa and 975 °C. The reconstructed precursor is a high-Al enstatite that formed at higher pressure and temperature conditions of ∼6 GPa and 1750 °C. The megacryst shows evidence for disintegrating into granular peridotite, and garnet and orthopyroxene within the granular peridotite are texturally and chemically identical to equivalent phases in the megacryst. Collectively, this evidence supports a common origin for the granular and exsolved portions of the sample. We hypothesize that high-Al enstatite was a common phase in the C-SCLM and that exsolution during cooling and stabilization of the C-SCLM could be the origin of most subcalcic garnets in depleted peridotites.