Objectives: To identify and synthesize evidence on the diagnostic accuracy of FENO for asthma in adults. Materials and Methods: Systematic searches (nine key biomedical databases and trial registers) were carried out on November 2014. Records were included if they recruited patients with the symptoms of asthma; used a single set of inclusion criteria; measured FENO50 in accordance with American Thoracic Society guidelines, 2005 (off-line excluded); reported/allowed calculation of true-positive, true-negative, false-positive and false-negative patients as classified against any reference standard. Study quality was assessed using QUADAS II. Meta-analysis was planned where clinical study heterogeneity allowed. Rule-in and rule-out uses of FENO were considered. Results: A total of 4861 records were identified originally and 1312 in an update. Twenty-seven studies were included. Heterogeneity precluded meta-analysis. Results varied even within subgroups of studies. Cut-off values for the best sum of sensitivity and specificity varied from 12 to 55 p.p.b., but did not produce high accuracy. 100% sensitivity or 100% specificity was reported by some studies indicating potential use as a rule-in or rule-out strategy. Conclusions and Clinical Relevance: FENO50 had variable diagnostic accuracy even within subgroups of studies with similar characteristics. Diagnostic accuracy, optimal cut-off values and best position for FENO50 within a pathway remain poorly evidenced.