Background: Will introducing a new diagnostic test benefit patient health? This question underpins the evaluation of diagnostic tests. Randomised controlled trials (RCTs), that randomise patients to diagnostic strategies and evaluate patient outcomes after the implementation of subsequent treatment, are the methodological goldstandard. However RCTs are only as informative as the outcomes they measure. This choice is particularly tricky in test-treat trials, since outcomes must evaluate the effects of a whole care pathway: diagnosis, consequent decision-making and treatment. Outcomes that fail to capture the effects of these complex interventions are likely to producemisrepresentative evaluations of performance. Objectives: To evaluate the methodological validity of outcome selection in test-treat RCTs. Methods: Published test-treat RCTs were identified through the Cochrane Central Register of Controlled Trials (Issue 2 2009, years 2004 7). Included RCTs evaluated test-treat strategies of any diagnostic modality and assessed patient outcomes after treatment. A new framework that explains the mechanisms by which tests might benefit patient health is used to analyse each test-treat comparison, hypothesise which outcomes should be measured to fully capture all putative benefits of a strategy, and extract and examine the outcomes that have actually been measured. Results: Of 12,619 titles, 106 (0.8%) articles reported RCTs of test-treat interventions evaluating at least one patient outcome. A wide range of clinical settings and testing modalities are represented, the most common being cardiovascular medicine (33%) and imaging (39%) respectively. A summary of outcomes measured is presented alongside an appraisal of whether they capture the theoretical processes of change as identified by the framework. Conclusion: Comment is given on the degree to which outcomes measured in test-treat RCTs capture the likely processes of change to patient health offered by the new strategy. The presentation will conclude with thoughts on how frequently published evaluations may be misrepresenting the performance of test-treat interventions.