Although several studies have examined the repeatability of objective refraction, data concerning the repeatability of subjective refraction under masked conditions, i.e., where the examiner is unaware of the refractive results, are limited. Accordingly, the present study compared the variability of both subjective and objective refractive techniques. Refractive error was measured in 12 subjects on 5 separate occasions. Conventional subjective procedures were used, with the exception that the sphere power scale on the phoropter was covered so that the examiner was unaware of the final result. Objective measurements were obtained using a Canon Autoref R-1 infrared autorefractor. The standard deviation (SD) of the five examinations was calculated for each individual and the mean values for the population sample determined. The mean SD's for the subjective and objective techniques were +/- 0.14 and +/- 0.18 D, indicating 95% confidence limits of +/- 0.27 and +/- 0.35 D, respectively. It is concluded that with either assessment technique, a change in refractive error of at least +/- 0.50 D should be adopted as the minimum significant shift in refractive status.