Because interpreting affords only limited opportunity for restatement or corrections, it can be seen as the practitioner's default version, with written translation representing a more polished rendition. Thus, a comparison of the target texts of interpreting and translation can shed light not only on the differences between the two modalities as such, but on the processes involved in each of the two as well. In this case, target texts from interpreting and translation were used to investigate cognate status, performance on false cognates, and cognate processing. In the first stage of this experiment, seven professional translators/interpreters interpreted an English text into their L1, Hebrew; four years later, they rendered the same ST in writing. The source text contained 51 words for which Hebrew offered true cognates and 10 for which it offered false cognates. The data show that: (a) cognate status is most often consistent across modalities; (b) noncognate synonyms are more prevalent in translation than in interpreting; (c) when a participant produces a noncognate translation in one modality and a cognate translation in the other, the noncognate is much likelier to be produced in translation, and (d) performance on false cognates is far superior in translation.