Darwin (1920) noted that when radiation can be neglected it should be possible to eliminate the radiation degrees-of-freedom from the action of classical electrodynamics and keep the discrete particle degrees-of-freedom only. Darwin derived his well known Lagrangian by series expansion in v/c keeping terms up to order (v/c) 2 . Since radiation is due to acceleration the assumption of low speed should not be necessary. A Lagrangian is suggested that neglects radiation without assuming low speed. It cures deficiencies of the Darwin Lagrangian in the ultra-relativistic regime.PACS numbers: 03.50. De, 11.10.Ef When radiation can be neglected the Lagrangian of classical electrodynamics, putting β = v/c, can be written,(1) In 1920 Darwin [1] expanded the Liénard-Wiechert potentials to second order in β = v/c and thus found that,and (hats are used for unit vectors),give the correct Lagrangian to this order. More recent derivations can be found in a few textbooks [2][3][4]. In particular Jackson [4] notes that using the Coulomb gauge (∇ · A = 0) makes the electrostatic Coulomb potential φ exact and moves all approximation to the vector potential A which obeys the inhomogeneous wave equation with the transverse (divergence free) current as source. [7]), and it is useful in various fundamental studies of electrodynamics [8][9][10][11]. The magnetic interaction described by the Darwin Lagrangian is essential in relativistic many-electron calculations as noted by Breit and others [12][13][14][15]. It has found applications in nuclear physics [16,17], and especially in plasma physics, for numerical simulation [18][19][20][21][22], thermodynamics and kinetics [23][24][25][26][27], as well fundamental theory [28][29][30]. Barcons and Lapiedra [31] noted that the Darwin approach is not valid for a relativistic plasma and therefore used a different approach to its statistical mechanics.Corrections to the Darwin Lagrangian have been discussed. Since a system of particles with identical charge to mass ratio does not dipole radiate a higher order expansion should be meaningful for such systems [32][33][34]. To that order, however, acceleration inevitably enters and must be handled in some way. Others have argued that since radiation is due to acceleration, v/c expansion is irrelevant, and further that radiation can be negligible even if the particle speeds are considerable (Trubnikov and Kosachev [35], Frejlak [36]). We will pursue that lead here.One frequently encounters the statement that the Darwin approach neglects retardation. This may be due to the fact that the, nowadays best known, elegant derivation by Jackson [4] hides the complications due to retardation. Nevertheless it is wrong. The derivations by Darwin [1] and by Landau and Lifshitz [2] show that the contribution of retardation to the Coulomb potential in the Lorenz gauge, is quite large. The main acceleration dependent part, however, vanishes either, as in Darwin's derivation, because it gives a total time derivative term in the Lagrangian, or, as in Landau ...