It has been suggested that the appearance of directional asymmetry and/or a reduced time constant of optokinetic afternystagmus (OKAN) might be a clinical index of vestibular imbalance. However, we do not know the limits for OKAN parameters in normal humans. Accordingly, we studied OKAN in 30 normal subjects using a "sampling" method, in which a number of values of OKAN are obtained by turning out the lights periodically during optokinetic stimulation. We found that the initial velocity of OKAN has a large intrasubject variability. Accordingly, if precision is desired so as to obtain 95% confidence that the measured mean of the initial velocity of OKAN is within 25% of the true mean in an individual subject, at least eight measurements of the initial OKAN velocity must be taken. When 12 measurements are made, all subjects had a minimum value of 5 degrees/s initial OKAN, and there was little directional asymmetry (mean of -0.47 degree/s +/- 3.13 degrees/s). The intrasubject variability of the time constant of OKAN was similar to the variability of initial OKAN velocity. However, because it is not possible to obtain repeated measures of the time constant in a short period of time, the time constant of OKAN is less likely to be useful in clinical testing.