The conductance method employs a multi-electrode catheter to generate an electrical field and measure intracavitary segmental conductances. Left ventricular (LV) volumes are calculated using an algorithm which assumes the electrical field to be homogeneous. This assumption may be violated leading to a non-linear relation between conductance-derived and true volumes. In addition, this relation may vary between segments. A new method is introduced which uses a more homogeneous field. Volume estimates using the conventional single excitation and the new dual excitation method were compared in a mathematical model of a canine LV, which was varied over a large volume range. With single excitation the slope factors, relating conductance-derived and true volumes, varied from 0.50 to 0.76 between segments and was 0.65 for total LV volume. Using dual excitation the segmental slope variability was reduced (range: 0.74-0.77) and the slope factor for total volume increased to 0.76. The linearity of the relation between conductance-derived and true volume was improved with dual excitation and extended over a larger range.