Background: Independent evaluation of clinical evidence is advocated in evidence-based practice. However, readers often rely on authors’ conclusions for quick summary messages. Cochrane systematic reviews (CSRs) are widely considered a reliable source of clinical information. Objectives: To assess whether a group of medical students and hospital staff could accurately derive their own conclusions of selected CSRs; to assess how these were influenced by their prior beliefs. Methods: Participants included hospital staff attending an EBP seminar (June 2008) and final-year medical students (November 2008). We showed the participants four CSR abstracts without conclusions. For each review, participants chose a conclusion for one specified outcome from six pre-defined options comprising different combinations of direction of effect and strength of evidence. We determined the best conclusion for each review by referring to the actual authors’ conclusions. Two chosen reviews concluded the intervention as beneficial (positive) and two non-beneficial (negative). Results: Participants (n=130) included 44 clinicians, 51 allied health staff (AHS) and 35 medical students. Each participant interpreted four reviews, making 520 ‘encounters’ in total. Overall, 62% correctly identified the direction of effect, and 30% chose the best conclusions by correctly identifying both the direction of effect and strength of evidence. More medical students (48%) chose the best conclusions compared to clinicians (22%) and AHS (23%) (p < 0.001). About half (47%) correctly identified the direction of effect against their beliefs. Positive reviews were more likely than negative reviews to change readers’ beliefs on the effect of the intervention (RR 1.82, 95% CI 1.29 to 2.56) and ‘convert’ those who were previously unsure (RR 1.94, 95% CI 1.34 to 2.80). Conclusions: Majority of the participants could not interpret CSR results accurately, with particular difficulty judging the strength of evidence. Positive reviews had greater impact than negative reviews on changing participants’ prior belief. Authors’ conclusions are still needed to guide readers in identifying appropriate clinical messages.