The reaction γ + d → p + n is studied at photon energies in the range 0.1 to 0.7 GeV. The approach is based on a non-covariant description of the deuteron (d) (accounting for the pn S- and D-components as well as for the virtual ΔΔ component) and a plane-wave description of the final pn state. The scattering amplitude incorporates, in addition to the impulse term, all the relevant one-pion exchange current terms corresponding to the Born and the rescattering terms in the process γN → N + π. We discuss contributions to the latter process induced by the N 33 ∗ (Δ) resonance as well as by the N 33 ∗ (S 11 , P 11 , D 13 ) resonances in the second bump. We consider two treatments of the N 33 ∗ contribution based on the isobaric model and a dispersion theory model. Predictions are presented for differential cross sections and for proton polarizations. We assess the roles of the various terms in the scattering amplitude. Some refinements are considered occasionally, bearing on final-state interaction effects and the ρ-meson exchange terms. We also examine in detail the sensitivity of predictions to various inputs and approximations, the most relevant being the range parameters for the πNN vertex, the off-shell extrapolation of the amplitude γN → πN and the retardation corrections. Recognizing that the indefiniteness in certain inputs may be the cause of appreciable uncertainties, we are led to seek for plausible prescriptions for these inputs by referring to measurements. We find that instrumental items, in order to achieve acceptable fits to differential cross sections, are a range parameter for the πNN vertex around 700–900 MeV/ c and an energy-dependent N 33 ∗ width. The predictions for the proton polarizations feature a much stronger sensitivity. In particular, we fail to explain the large enhancement observed at around 500 MeV photon energy.