We propose and test a pair potential that is accurate at all relevant distances and simple enough for use in large-scale computer simulations. A combination of the Rydberg potential from spectroscopy and the London inverse-sixth-power energy, the proposed form fits spectroscopically determined potentials better than the Morse, Varnshi, and Hulburt-Hirschfelder potentials and much better than the Lennard-Jones and harmonic potentials. At long distances, it goes smoothly to the correct London force appropriate for gases and preserves van der Waals's "continuity of the gas and liquid states," which is routinely violated by coefficients assigned to the Lennard-Jones 6-12 form.There are at least three classes of interatomic potentials. Commercial codes use the Lennard-Jonesand harmonicforms, which are accurate near the minimum at r = r 0 . But this first class of potentials may be too simple for complex materials away from from equilibrium.(β = (κr 2 0 − 1)/(2r 2 0 )), and Hulburt-Hirschfelder [3][4]) potentials represent a second class of potentials [4] accurate over a wider range of distances. Quantum chemists [5,6,7,8,9] have derived a third class that reproduce spectroscopic and thermodynamic data with impressive fidelity. But the potentials of this class involve many parameters and may be too cumbersome for use in large-scale simulations.We propose and test a formthat is nearly as accurate as the class-3 potentials but simpler than many class-2 potentials. It is a combination of the Rydberg formula used in spectroscopy and , and e = 47.6Å 12 ) (solid, red) fits the RKR spectral points for the ground state of molecular hydrogen (pluses, blue) and gives the correct London tail for r > 3Å. The Lennard-Jones VLJ (dashes, green) and harmonic VH (dots, magenta) forms fit only near the minimum.the London formula for pairs of atoms. In Eq.(6), the terms involving a, b, and c were proposed by Rydberg [10] to incorporate spectroscopic data, but were largely ignored until recently [11,12]. The constant d = C 6 is the coefficient of the London tail. The new term e r −6 cures the London singularity. As r → 0, V (r) → a, finite; as r → ∞, V (r) approaches the London term, V (r) → −d/r 6 = −C 6 /r 6 . In a perturbative analysis [13], the a, b, c terms arise in first order, and the d term in second order.To test whether the hybrid V (r) can represent covalent bonds far from equilibrium, we used Gnuplot [14] to fit Eq.(6) to empirical potentials for molecular H 2 , N 2 , and