In 1985 Fischer gave a simple explanation for the behaviour of the E-polarization magnetotelluric phase over a 2-D dike. This phase behaviour is a direct consequence of the rearrangement of the current flow in a 2-D structure as opposed to the flow in a simple uniform half-space. In spite of its usefulness it is shown that the simple phase rule is only qualitatively correct. For B-polarization it seemed at first that the simple phase rule would not remain valid. However, it is found that if the very different current distributions for E -and B-polarization are correctly interpreted, the simple phase rule retains its validity qualitatively. Arguments are given for the very general validity of this rule at the surface of any 2-D structure, even in the presence of complicated topography. The rule is best formulated in terms of two statements: (a) the magnetotelluric phase is a continuous function over any structure, even across outcropping resistivity contrasts; and (b) where the current is drawn to greater depths the phase will rise, and where the current is concentrated near the surface the phase will drop.