BACKGROUND : Obstructive sleep apnea syndrome (OSAS) is a common disorder with significant morbidity and mortality. We aimed to evaluate the predictive accuracy of the Berlin questionnaire in patients with suspected OSAS undergoing PSG in the sleep laboratory setting against those going through the Embletta™ portable diagnostic system (Embletta PDS) at home. METHODS : Patients with suspected OSAS were recruited from respiratory clinics to complete Berlin questionnaire and Epworth Sleepiness Score (ESS). Patients were randomized to undergo either home-based sleep test (group A) or hospital-based polysomnography (PSG) (group B). RESULTS : Three hundreds and sixteen subjects with newly referred suspected OSAS were recruited and randomized into group A (n = 157) and group B (n = 159). The prevalence of moderate to severe OSAS defined as apnea-hypopnea index (AHI) ≥ 15/h was 54%. The Berlin questionnaire identified 69.7% (n = 99) of subjects as high risk in group A and 77.5% (n = 100) in group B. The sensitivity, specificity, negative predictive value (NPV), and positive predictive value (PPV) of the questionnaire to predict an AHI ≥ 15/h as diagnosed by PSG was 78, 23, 67 and 35%. When compared with Embletta PDS, the specificity and NPV increased to 48 and 63%. The area under the Receiver Operator Curve (ROC) based on PSG (AUC = 0.539, 95%CI 0.417, 0.661) and based on home Embletta (AUC = 0.712, 95%CI 0.617, 0.907). CONCLUSIONS : The questionnaire was not reliable in predicting OSAS through PSG AHI whereas there was some predictive ability in discriminating patients with OSAS from normal subjects based on home Embletta sleep test. BACKGROUND : The study was registered at ClinicalTrials.gov (Identifier: NCT01828216) on 10 April 2013.