We present a model of steady-state transport of Xe through the upper mantle with inputs from an undergassed lower mantle by mass transport without fractionation and subducted atmospheric gases. The model is explored using box model mass transport equations for each isotope of Xe coupled with the transport equations for He. The mantle is assumed to have had initially uniform Pu,I, and U concentrations and is divided into two reservoirs. The lower mantle is assumed to have evolved approximately as a closed-system containing initially trapped Xe along with xenon isotopes produced by 129I, 244Pu, and 238U decay. The concentration and present isotopic composition of lower mantle Xe are unknowns. The upper mantle is considered to have inputs of Xe by mass transfer from the lower mantle at hotspots and from the atmosphere by subduction. In addition, the decay of U within the upper mantle contributes radiogenic Xe as well as 4He. The flows of Xe into the upper mantle are balanced by flows out of the upper mantle at mid-ocean ridges and hotspots, so that the upper mantle Xe concentration is assumed to be in a steady state. In this model, upper mantle Xe reflects the isotopic characteristics of the lower mantle altered by mixing with subducted atmospheric Xe and the present decay of U in the upper mantle. This is in contrast to previous models in which upper mantle Xe is what remains after degassing of the atmosphere, with a present isotopic composition that reflects the integrated history of continuous losses of Xe and grow-in of radiogenic Xe isotopes within the upper mantle. The observed correlation of 129Xe /130Xe and 136Xe /130Xe ratios for MORB found by other workers is interpreted here as reflecting different degrees of atmospheric contamination to a MORB Xe composition. Consideration of the production and transport equations yields the following conclusions: (1) radiogenic 136Xe in the lower mantle is dominantly from spontaneous fission of Pu; (2) radiogenic 136Xe in the upper mantle is composed of radiogenic 136Xe from the lower mantle substantially augmented by spontaneous fission of U in the upper mantle; (3) the concentration of Xe in the lower mantle is < 10 -2 that of carbonaceous chondrites; (4) radiogenic 129Xe is stored in the lower mantle, and the 129Xe excesses seen in the upper mantle are due to mass transfer of Xe from the lower mantle to the upper mantle; (5) the residence time of Xe and He in the upper mantle is ∼1.4 Ga; (6) a substantial portion of nonradiogenic Xe in the upper mantle may be accounted for by subduction of atmospheric Xe; (7) if a large fraction of the total Xe in the upper mantle is subducted atmospheric Xe, then the lower mantle 129Xe /130Xe and 136Xe /130Xe ratios Xe must be large; and (8) the ratio of radiogenic 129Xe to 127I in the lower mantle is about 10 -2 times that seen in meteorites. This indicates that Xe was lost from the Earth ∼-10 8 years after solar system formation. This occurred either from the fully formed Earth or from Earth-forming materials that accreted late. A consequence of the model is that most of the atmosphere was derived from a Xe reservoir with distinct radiogenic isotope characteristics and which is not presently represented within the Earth. In addition to the radiogenic 136Xe and 129Xe contributed to the atmosphere from the upper mantle, Xe from a gas-rich component with Xe/I ratios higher than carbonaceous chondrites is required. One possible source of this component is gas-rich cometary material. If atmospheric Xe has been subject to severe fractionation of the nonradiogenic isotopes due to atmospheric losses to space, as proposed by others, then lower mantle nonradiogenic xenon isotopes will be fractionated with respect to the atmosphere. The extent to which such fractionation is exhibited by upper mantle Xe depends upon the proportion of subducted Xe to lower mantle Xe present in the upper mantle.