Healthy human subjects, when roll tilted in darkness, make systematic errors in estimating subjective visual vertical (SVV). Typically, roll tilt underestimation occurs at angles beyond 60 degrees (A-effect). At smaller tilt angles, overestimation may occur (E-effect). At approximately 135 degrees whole-body roll tilt, Kaptein and Van Gisbergen (2004, 2005) found an abrupt A/E transition, the exact location of which depended on the preceding rotation direction indicating hysteresis. Since this was observed using relatively fast roll velocity, it remains unclear whether the described hysteresis is dynamic or static. To clarify this uncertainty, we continuously rotated nine healthy subjects about the earth-horizontal naso-occipital axis, while they performed SVV adjustments every 2 s. Starting from the upright position, three full quasi-static constant velocity rotations (2 degrees/s) were completed in both directions (CW: clockwise; CCW: counterclockwise). SVV deviation from earth-verticality was plotted as a function of whole-body roll position. A bimodal Gaussian distribution function was fitted to SVV differences between CW and CCW rotations. A-effects (peaks at 88 degrees and 257 degrees chair position) at identical whole-body positions were larger after rotations from upside-down than after rotations from upright (average peak difference: 26 degrees). These results demonstrate static hysteresis for SVV estimation.