Alexander's law (AL) states that the slow-phase velocity of spontaneous nystagmus of peripheral vestibular origin is dependent on horizontal gaze position, with greater velocity when gaze is directed in the fast-phase direction. AL is thought to be a compensatory reaction resulting from adaptive changes in the horizontal ocular motor neural integrator. Until now, only horizontal eye movements have been investigated with respect to AL. Because spontaneous nystagmus usually includes vertical and torsional components, we asked whether horizontal gaze changes would have an effect on the 3D drift of spontaneous nystagmus and, thus, on the vertical/torsional neural integrator. We hypothesized that AL reduces all nystagmus components proportionally. Moreover, we questioned the classical theory of a single bilaterally organized horizontal integrator and searched for nonlinearities of AL implying a network of multiple integrators. Using dual scleral search coils, we measured AL in 17 patients with spontaneous nystagmus. Patients followed a pulsed laser dot at eye level jumping in 5 degrees steps along the horizontal meridian between 25 degrees right and left in otherwise complete darkness. AL was observed in 15 of 17 patients. Whereas individual patients typically showed a change of 3D-drift direction at different horizontal eye positions, the average change in direction was not different from zero. The strength of AL (= rate of change of total velocity with gaze position) correlated with nystagmus slow-phase velocity (Spearman's rho = 0.5; p < 0.05) and, on average, did not change the 3D nystagmus drift direction. In general, eye velocity did not vary linearly with eye position. Rather, there was a stronger dependence of velocity on horizontal position when subjects looked in the slow-phase direction compared to the fast-phase direction. We conclude that the theory of a simple leak of a single horizontal neural integrator is not sufficient to explain all aspects of AL.