Physical aging below the glass transition temperature, T-g, is generally understood to be governed by the same slow degrees of freedom that are responsible for structural relaxation in the equilibrium state above T-g. Provided a reliable model of aging is at hand, it should thus be possible to extract very long relaxation time constants from experimental data on physical aging. Two very different models of aging are investigated in this respect using data for various molecular glass-forming liquids extending out to aging times of 3 X 10(6) s. It turns out that application of the well-known KAHR or TNM model does not provide a significant advantage over a recently proposed phenomenological approach that is much simpler.