We consider the vector-dominance model applied to single-pion photoproduction. Dispersion relations are written for a set of invariant amplitudes. Particular attention is paid to the problems of gauge invariance and kinematic singularities. At high energies our results coincide, to leading order only, with the use ofs-channel helicity amplitudes. We argue that the use of the Donohue-Hogaasen frame is meaningless. Finally we discuss the ambiguity in the choice of best extrapolation path.