We give several parametrizations for the elastic scattering and radiative capture cross sections for low neutron bombarding energy and discuss the relationship between the corresponding resonance parameters. We then perform an extensive investigation of the valence radiative capture model of Lane and Lynn. This model is formulated here in the frame of the shell-model approach. We exhibit the similarities and differences between our results and those derived from the R-matrix approach by Lane and Lynn on the one hand and from the optical-model approach by Lane and Mughabghab on the other hand. Particular attention is paid to the choice of the average potential well in the shell-model approach, in relation to the proper way to identify theoretical quantities and phenomenological parameters. We show that practically equivalent results can be obtained from a complex average potential well and from a suitably chosen real potential well, respectively. The following topics are investigated formally and numerically: dependence of the various theoretical expressions on the choice of the (real or complex) average potential well; relative importance of external and internal capture; dependence of photon widths and background cross section on mass number (for thermal energy and for E = 100 keV); dependence of the resonance parameters and background cross sections on energy, for A = 60; comparison between experimental data and theoretical values for radiative capture on 56Fe and 60Ni. We discuss the conditions of validity of the valence capture model. In particular, we investigate the role of the giant dipole resonance and of the closed channels. We argue that the success of the valence capture model is intimately related to the importance of external capture. The contribution of the low-lying excited target states is investigated formally and numerically; it increases with mass number and tends to diminish the correlation between neutron and photon widths, which is implied by the valence capture model.