The validity of Listing's law was reinvestigated by means of a direct test. Horizontal, vertical and torsional eye movements were measured simultaneously with a recently developed scleral induction coil. Either eye of 4 subjects was measured monocularly. Eye position were measured in Fick coordinates and ocular torsion values were compared to the theoretical ones predicted by Listing's law. During consecutive measurements in the primary position torsion values were close to zero although considerable fluctuations of torsion were seen. Torsion values in the secondary positions were also close to zero. In the tertiary positions torsion in the direction as predicted by Listing's law and increasing with eccentricity was recorded. In the temporal quadrants mean torsion was quantitatively in agreement with Listing's law; torsion values in the nasal quadrants however showed systematically larger values and this discrepancy increased with eccentricity to more than 50%. Statistical support for this finding however, was seen only in 4 out of 8 eyes. Symmetry could be obtained by shifting the chosen horizontal primary position (gaze parallel to the midplane) in the temporal direction; as a consequence all measured torsion values would exceed the ones specified by Listing's law. Torsion values varied idiosyncratically among subjects and among the left and right eyes of any one subject. It is concluded that Listing's law specifies ocular torsion only approximately: physiological eye movements show considerable stochastical as well as systematical deviations from this law.