A proposal has been made by Ferguson and Hudson (fur. Polym. J., 29 (1993) 141-147; J. Non-Newtonian Fluid Mech., 52 (1994) 121-135) that three-dimensional representation of extensional flow data can be used to resolve apparent disagreements among the results from a variety of extensional flow experiments. A theoretical investigation of the procedure, which involves plotting a transient extensional viscosity against strain and time is carried out in this paper. We then seek to draw some conclusions about the validity and limitations of the approach. The method does not work for purely viscous non-newtonian liquids or for simple (co-rotational and upper convected) Maxwell models. However, the failure of results to lie close to a unique surface (for any particular material) is most marked in situations where our theoretical models are least reliable. More work, both experimental and theoretical, is required.