The metallization lattice constant for fee Xe has been calculated in three different, independent ways. The result is relatively insensitive to basis set, details of local exchangecorrelation potential, or use of the muffin-tin potential. The calculated metallization lattice constant is 7.9 a.u. or P=1. 28 Mbar, confirming Ross and McMahan's calculation but disagreeing with Nelson and Ruoff's experimental value. PACS numbers: 71.30.+ h, 64.30.+ t, 71.25.Tn Metallic conduction in Xe at 330 kbar and 32 K has been reported by Nelson and Ruoff. 1 Subsequently, Ross and McMahan 2 reported an augmented-plane-wave calculation (APW) with HedinLundquist local exchange and correlation. They found no metallization until P= 1.3 Mbar, a result confirmed by Worth and Trickey 3 and by Wilkins and Williams. 3 In view of the fact that simple local-density theories always give a smaller band gap at P = 0 than the experimental value, 4 it is quite surprising that the calculated metallization pressure is almost a factor of 4 greater than that reported from experiment. It is especially surprising in view of the excellent agreement 2 ' 5 between calculated and measured P-V curves over the entire pressure range. We have therefore undertaken an extensive check to see whether we could uncover any flaw in the pressure calculations.Our first concern was muffin-tin effects introduced by the standard implementation of the APW formulism, as well as related effects in the newly developed augmented-spherical-wave method 6 used by Wilkins and Williams (Ref. 3). Recently two of us (A.K.R. and S.B.T.) have developed a computer code which enables one to use our selfconsistent, muffin-tin APW output as input to the Wang-Calloway 7 linear combination of Gaussiantype orbitals (LCGTO) code. We are enabled thus to assess muffin-tin effects by direct, systematic calculations. The LCGTO Ansatz opens up the possibilty of basis-set inadequacies, of course. We have attempted to confront that problem in two ways. First, we have chosen a rather large Gaussian basis (16 s, 12 p, 8 d) of Huzinaga. 8 Second, we have used the local orbitals, mixed basis (Slater-type orbitals plus plane waves, hereafter STO + PW) scheme of Kunz. 9 Here we used STO basis set contracted to 5s, 4/>, 2d functions plus 113 plane waves. Our third concern was with the details of the local-exchange-correlation model employed. To probe this possible source of discrepancy with experiment, we did the APW and LCGTO calculations usingXa exchange-correlation with the so-called virial-theorem a, 10 while the STO + PW calculation used the KohnSham-Gaspar (KSG) value of a plus a self-energy correction. 11 All calculations were done on a 256-point first-Brillouin-zone mesh c Neither the LCGTO nor the STO + PW computer codes are equipped at present to compute total energies and hence T=0 K static lattice P-V curves. Therefore, we have tabulated our data as a function of the cubic lattice constant and used the APW-Xa equation of state to convert to a corresponding pressure (see Table I). Since it is...