Six experiments used an illusory words paradigm to demonstrate that repetit
ion blindness (RB) in orthographically similar words affects only the words
' shared letters. Rapid serial visual presentation streams of words and wor
d fragments allowed the unique letters of the 2nd critical word to combine
with a subsequent fragment to create a word, as in rock shock ell. The illu
sory word shell was reported 2-3 times as frequently in RE conditions as in
control conditions. Further experiments ruled out letter migration, contou
r summation, and differences in processing load as explanations for the res
ults. These findings are inconsistent with current proposals that orthograp
hic RE represents similarity inhibition or lexical competition or that it r
eflects problems with word-level token individuation.