Objective: Back translation is considered by many to be an integral part of the translation methodology for PRO documents. But while the ISPOR Task Force paper (2005) describes it as a critical component in the translation process without which final translations risk containing undetected different content from the source document, it is criticised by McKenna et. al. (2005) for having “no clear scientific basis” and for leading to potentially misleading impressions of the translation. In contrast to the more conceptual forward translation, using back translations as a tool in the translation of PRO documents usually requires a more literal translation to provide an accurate “window” into the translation. This research compares literal with non-literal back translations in PRO documents and aims to ascertain the circumstances in which each type of back translation would be most appropriate. Methods: A selection of PRO translation projects has been reviewed based on the extent to which the back translation was literal or not. Results: Some literal translations can be confusing and un-natural sounding while others suggest the forward translation is incorrect where this is not necessarily the case. Examples of both types of back translation include the literal back translation of a Turkish phrase as “did your head turn …over the last 24 hours?” This makes little sense because the forward translation was idiomatic. Back translations of the Ukrainian for “please tick one box” resulted in “please tick one answer” (conceptual) and “please tick one square” (literal) – here the literal translation is closer. The literal back translation of the Czech for “seizure” shows an error in the Czech where the conceptual back translation does not highlight this. Conclusion: Literal back translations can be misleading in idiomatic phrases but are usually more beneficial for symptoms and health states.