In the present study, we examined error detection/correction during collaborative writing. Subjects were asked to identify and correct errors in two contexts: a passage written by the subject (familiar text) and a passage written by a person other than the subject (unfamiliar text). A computer program inserted errors in function words prior to the proofreading of both passages. The errors preserved the lexical status of the original words, making the errors difficult to detect by commercially available spell-checking functions. The two passages permitted us to ask whether unfamiliar errors (i.e., errors not committed by the proofreader) would be easier to detect/correct in a familiar text than in an unfamiliar text. It was hypothesized that the proofreader's substantial familiarity with his/her own writing might make the unfamiliar errors salient, thereby enhancing error detection. Alternatively, this elevated familiarity could impair error detection by increasing the predictability of the text being proofread and thus the likelihood of overlooking errors. The results provided support for the "perceptual salience" account and suggested that an environment that diminishes the proofreaders' familiarity with their own writing (e.g., substantive typescript changes) can also diminish the salience of the errors committed by others.