Classical assumptions concerning the contribution of the solvent to the viscoelastic properties of polymer solutions are shown to be invalid. The presence of polymer affects the mean solvent rotational mobility, in some cases even enhancing it relative to the neat solvent case. The resulting changes in local friction modify both the polymer relaxation times and the solvent contribution to the solution properties. The effect is quantified through an intrinsic "effective solvent viscosity", [ηe], which can be positive or negative. To a first approximation, [ηe] appears as an additive correction to the measured intrinsic viscosity, [η]; this correction can be substantial, particularly for polymers of low molecular weight. Furthermore, [ηe] largely accounts for the anomalous high-frequency limiting viscosity, designated [η′∞], even in those cases where [η′∞] is negative. These results remain to be incorporated successfully into a kinetic theory framework; however, a phenomenological interpretation is offered. The central, unresolved issue is to predict how the local dynamics of molecules in a mixture are modified relative to the unmixed state; the same issue arises in the dynamics of polymer blends.