Test-retest data were obtained binocularly, monocularly and monocularl
y in the presence of glare from 20 healthy older patients using the Pe
lli-Robson and Regan (96% and 11% contrast) charts. Reliability was as
sessed by means of two indices: intra-class correlation coefficient an
d coefficient of repeatability which determines the 95% confidence lim
its for change. Binocular viewing showed the highest intra-class corre
lation and the smallest confidence interval with all the charts. Monoc
ular viewing in the presence of glare showed the lowest correlation an
d the largest confidence interval. Results suggest that when assessing
the significance of change, the confidence interval should be determi
ned for identical viewing conditions.