We update the determination of the isovector nucleon electromagnetic self-energy, valid to leading order in QED. A technical oversight in the literature concerning the elastic contribution to Cottingham's formula is corrected and modern knowledge of the structure functions is used to precisely determine the inelastic contribution. We find δM γ p−n = 1.30(03)(47) MeV. The largest uncertainty arises from a subtraction term required in the dispersive analysis, which can be related to the isovector magnetic polarizability. With plausible model assumptions, we can combine our calculation with additional input from lattice QCD to constrain this polarizability as: βp−n = −0.87(85) × 10 −4 fm 3 .PACS numbers: 13.40. Dk, 13.40.Ks, 13.60.Fz, 14.20.Dh Given only electrostatic forces, one would predict that the proton is more massive than the neutron but the opposite actually occurs [1-3]:Before we knew of quarks and gluons there were many attempts to explain this contradiction, see Ref.[4] for a review. We now know there are two sources of isospin breaking in the standard model, the masses of the up and down quarks as well as the electromagnetic interactions between quarks governed by the charge operator. The effects of the mass difference between down and up quarks are larger and of the opposite sign than those of electromagnetic effects, see the reviews [5][6][7]. The net result of the quark mass difference and electromagnetic effects is well known, Eq. (1), but our ability to disentangle the contributions from these two sources remains poorly constrained. In contrast, lattice QCD calculations have matured significantly. There are now calculations performed with the light quark masses at or near their physical values [8][9][10][11][12], reproducing the ground state hadron spectrum within a few percent. These advances have allowed for calculations to begin including explicit isospin breaking effects from both the quark masses [13][14][15][16][17] and electromagnetism [15,[18][19][20][21]. While the lattice calculations of m d − m u effects are robust, the contributions from electromagnetism are less mature and suffer from larger systematics, due in large part to the disparity between the photon mass and a typical hadronic scale. Improved knowledge of m d − m u and its effects in nucleons will enhance the ability to use effective field theory to compute a variety of isospin-violating (charge asymmetric) effects in nuclear reactions [7,[22][23][24][25][26][27]].An application [28] of the Cottingham sum rule [29], which relates the electromagnetic self-energy of the nucleon to measured elastic and inelastic cross sections, gives the result δM γ p−n = 0.76 ± 0.30 MeV. Given the high present interest in the precise value of δM γ p−n and its many possible implications, it is worthwhile to revisit this result. Many high quality electron scattering experiments have been performed since 1975 and there have also been theoretical advances. The central aim of this work is to provide a modern, robust evaluation of δM γ p−n . We wil...