Prior theories involving electroosmotic, electrokinetics, and other effects have proved inadequate to explain the observed sample spreading and performance degradation in continuous flow electrophoresis (CFE). We suggest electrohydrodynamic flows as the main cause. These findings should contribute to efforts to improve the performance of CFE. It is shown theoretically that an electric field (AC or DC), perpendicular to a circular filament of conducting fluid surrounded by fluid of a different conductivity, produces an electrohydrodynamic flow, which distorts the filament into an ellipse. This distortion is similar to that described by G. I. Taylor for spherical drops. The major axis of the ellipse is either parallel to or normal to the field, depending on the conductivity and dielectric constant of the filament fluid relative to those of the surrounding fluid. For equal dielectric constants, the major axis is parallel to the field if the filament conductivity is greater than that of the surrounding fluid, and normal to the field otherwise. The flow and distortion rate is proportional to the square of the applied field. It is further shown theoretically that the flow associated with an elliptic cross section maintains the elliptic shape while it continues the distortion, provided the deviation from a circular cross section is small. As the ellipse stretches, small deviations from an elliptic cross section appear. It is shown, using an energy argument based on the assumption of an elliptic cross section, that the circular filament continues to flatten indefinitely, forming a ribbon, either parallel to the field or normal to it. The nature of the analysis, the physics of the flow, and the related experiments appear to confirm this behavior. For our experiments, we used an aqueous electrolyte (barbital buffer) and a sample of the same material with polystyrene latex added for visibility, in a CFE-type apparatus. The flow rate and configuration were typical of CFE. Electrokinetic and electrophoretic effects, and electroosmosis, were eliminated by using an AC field. Distinctive ribbons were formed, in both directions, at small fields of order 50 V/cm or less. All observations were consistent with the theories described above.