It is proposed that noble gases in the mantle reside mainly in a “hybrid pyroxenite” component that is formed as melt from old subducted oceanic crust reacts with surrounding refractory peridotite under mid-ocean ridges. The gases are presumed to have been acquired early in Earth history, and mixing under mid-ocean ridges of melts from pyroxenites and old crust would pass the noble gases to successive generations of pyroxenites. Much of the complement of noble gases would not be erupted at mid-ocean ridges and would recirculate in the mantle, so the hybrid pyroxenite would degas rather slowly over Earth history. The hybrid pyroxenite would also be denser than average mantle and would tend to settle into the D″ zone at the base of the mantle, in the same way as subducted oceanic crust. Because residence times in D″ are longer than in the rest of the mantle, the D″ pyroxenite would be less degassed and its noble gases would be less radiogenic than that in the mid-ocean ridge basalt (MORB) source. Plumes could therefore tap a mixture of old, degassed oceanic crust and less degassed hybrid pyroxenite. This could resolve the long-standing question of the source of unradiogenic helium in many ocean island basalts (OIBs). Abundances of noble gases are not very well constrained and are plausibly larger than conventional estimates, which would remove the need for a large, deep, “undegassed” reservoir. The ⁴⁰Ar mass balance may be satisfied if account is taken of uncertainties in the potassium content of the continental crust and the bulk silicate Earth and in the Earth's total budget of ⁴⁰Ar. A relatively simple quantitative theory for the evolution of the noble gases is presented that accounts for the present concentrations and isotopic compositions of helium, neon, and argon in the MORB and OIB source regions. In particular, it confirms that longer residence times in D″ can account for the less radiogenic noble gases in OIBs.