Abstract: With the introduction of the new RevMan5 software all Cochrane reviews will now assess the risk of bias of primary studies included in a review according to the guidelines described in the Cochrane handbook. Reviewers have to rate the adequacy of sequence generation, allocation concealment, blinding, completeness of outcome data and outcome reporting. The handbook gives detailed examples which help reviewers to make their decisions. In most instances this information will allow straightforward assessments. However, the handbook cannot cover all eventualities. For example, among a group of patient-blind studies one trial might include a test of the success of blinding indicating some degree of unblinding while the remaining trials were described as patient-blind, but do not report formal testing of blinding. Should reviewers consider the trial testing the success of blinding as having a high risk of bias while the remaining trials are deemed to have a low risk of bias? After an introductory lecture we will discuss examples for assessment difficulties from recent reviews. Participants are encouraged to bring own examples.