One disadvantage of using high-contrast letters as test objects when measuring visual acuity is the fact that they are not of equal legibility. A number of charts are now commercially available that assess contrast sensitivity using letter targets. This study attempted to assess the legibility of letters at contrast threshold on the Pelli-Robson letter contrast sensitivity chart by determining the percentage of correct responses for each of the ten Sloan letters at contrast threshold. Results of 493 contrast sensitivity measurements taken in optometric practice indicated that there is a definite difference in legibility between letters at contrast threshold as for letters at acuity threshold. The data suggest that the probability of correctly identifying two out of a group of three letters at threshold on the Pelli-Robson chart varies between 67% and 97% due to letter type alone. Because of the very regular and pronounced miscalling of the letter C as an O, we suggest that this should be accepted as a correct call during threshold measurements on the Pelli-Robson chart. This helps to balance the legibility of different groups of letters.