This study assessed the repeatability of the Nikon NRK-8000, the Nidek AR-1000, and subjective refraction. In addition, the accuracy of the Nikon and the Nidek were evaluated in comparison to subjective refraction. Measurements were taken with all 3 techniques on 2 separate occasions with a test-retest separation of at least 24 h. The right eyes of 30 normal subjects were used. Repeatability and accuracy statistics and plots were generated using matrix representations of dioptric power. Subjective refraction was the most repeatable method, with the coefficient of repeatability (COR) found to be 0.611, 0.224, and 0.490 in the vertical, torsional, and horizontal meridians. The autorefractors' COR was found to range from 0.712 to 0.826 for the vertical and horizontal meridians, whereas the torsional meridian ranged from 0.224 to 0.319.