OBJECTIVE: To provide a theoretical basis for the selection of the anode-cathode configuration in spinal cord stimulation for the management of pain when one percutaneous epidural electrode or two electrodes in parallel are used. METHODS: A computer model of spinal cord stimulation at T8-T9 was used to calculate the dorsal column areas recruited in stimulation by various configurations used in clinical practice. RESULTS: Tripolar (or bipolar) stimulation by a single electrode, symmetrically placed over the dorsal columns, recruits the largest area and will give the widest paresthesia coverage. Stimulation by two symmetrically placed electrodes connected in parallel to a single channel pulse generator may give similar results, because of their generally smaller distance from the spinal cord, but a ''summation effect'' does not exist. A smaller dorsal column area is activated when two offset electrodes are used. An electrode placed laterally or transverse bipolar stimulation results in unilateral, usually segmentary, paresthesia. CONCLUSIONS: The relative positions of cathodes and anodes and their distance from the spinal cord are the major determinants of dorsal column/dorsal root activation and paresthesia distribution. The large interpatient variability of the intraspinal geometry is the main cause of differences in paresthesia coverage among patients having optimally placed electrode(s). Changes of paresthesia coverage over time are more probable when multiple electrodes are used.