Contrast sensitivity using Vistech plates was measured repeatedly in normal subjects to determine predictive power. The contrast sensitivity measures were quite stable from the first administration, and practice effects were not observed. Test-retest reliability was lowest for the lower spatial frequencies, but was adequate throughout if more than one measurement was made. Threshold scores for all five spatial frequencies were generally not independent and factor analysis revealed that only two spatial frequency domains were relatively independent: high and low. The correlations between adjacent spatial frequencies were highest, and correlations decreased as distance between spatial frequencies increased. To detect subtle differences all three Vistech plates should be used to arrive at a threshold for each spatial frequency and, if time permits, more measures should be taken at the lower spatial frequencies to compensate for the lower reliability. This method is believed to be adequate for repeated-measures clinical and experimental studies.