Power counting is applied to relativistic mean-field energy functionals to estimate contributions to the energy from individual terms. New estimates for isovector, tensor, and gradient terms in finite nuclei are shown to be consistent with direct, high-quality fits. The estimates establish a hierarchy of model parameters and identify how many parameters are well constrained by bulk nuclear observables. We conclude that four (possibly five) isoscalar, nongradient parameters, one gradient parameter, and one isovector parameter are well determined by the usual bulk nuclear observables.Relativistic mean-field (RMF) descriptions of the nuclear many-body problem have existed for nearly fifty years [1]. An extensive body of work [2][3][4] has shown that RMF theories can realistically reproduce the bulk and single-particle properties of medium to heavy nuclei. Successful RMF theories are characterized by large, neutral, Lorentz scalar and vector potentials (roughly several hundred MeV at nuclear equilibrium density), a new saturation mechanism for nuclear matter that arises from relativistic (i.e., velocity-dependent) interaction effects from the scalar potential, and a nuclear spin-orbit force that is determined by this velocity dependence. The vector potential provides an efficient representation of the short-range nucleon-nucleon (NN) repulsion, while the scalar potential provides the midrange NN attraction by simulating correlated, scalar-isoscalar two-pion exchange, which is the most important pionic contribution for describing the bulk properties of nuclear matter.For many years, RMF model parameters were determined by fitting them to the properties of nuclei or nuclear matter using various "black-box" fitting schemes [5][6][7]. Direct connections between the RMF parameters and the systematics of nuclear observables have begun to emerge only recently [8][9][10][11]. The modern theoretical viewpoint underlying the RMF, which relies on the ideas of Effective Field Theory (EFT) and of Density Functional Theory (DFT) [11,4,12], provides a systematic truncation procedure for the RMF energy functional. However, it introduces more parameters than can be unambiguously determined by the relevant data. The present work uses EFT power counting to examine the quantity of information contained in the bulk and single-particle input data, and to identify how many parameters can be reliably determined by these data.The original Walecka model [13] (without nonlinear field interactions), contained two free isoscalar parameters in nuclear matter, which were used to fit the equilibrium density and binding energy. An isovector coupling was added [14] to reproduce the bulk symmetry energy, and for finite nuclei, a single scale parameter (the scalar "meson" mass) was chosen to reproduce the rms charge radius of 40 Ca. These procedures yielded reasonable results for charge-density distributions of spherical nuclei and for single-particle spectra, revealing that the nuclear shell model was obtained automatically without any specific adjust...