We studied excyclotorsion prospectively in 12 non-strabismic normal patients, 24 patients with unilateral superior oblique paresis (SOP), and 14 patients with bilateral SOP to determine which test and position, and what amount of torsion best discriminates between bilateral and unilateral SOP. For each patient, we measured torsion at near in primary position and in 20 degrees downgaze, using both the double Maddox rod (DMR) and Bagolini lens (BL) tests. We calculated the average of three measurements for each test in each position and the differences in mean excyclotorsion between downgaze and primary position for each patient for each test. To discriminate between unilateral SOP and normal, and bilateral and unilateral SOP patients, the best test and position combination was the DMR in downgaze (P = .0001). The probability of a torsion measurement indicating a unilateral SOP rather than a normal value was 95% at 6 degrees of excyclotorsion; of bilateral rather than unilateral SOP, 90% at 20 degrees. The range in the three torsion readings within patients with either DMR or BL in either position varied up to 7 degrees.