Background The STOP-BANG is a simple obstructive sleep apnea (OSA) screening tool, part questionnaire (STOP) and part demographic or physical measures (BANG), developed for use in preoperative surgical clinics. This study assessed sensitivity and specificity of the instrument among patients referred to a sleep disorders laboratory, and also its performance characteristics when BANG physical measures are patient-reported rather than measured. Methods Adults referred for diagnostic polysomnography completed the STOP questions and answered four yes/no questions (BANG self-reported) about their body mass index (weight and height), age, neck circumference, and gender, which were also assessed by laboratory technologists (BANG-measured). Results Among N=219 subjects (mean age 46.3 ± 13.9 [s.d.] years; 98 [44.8%] males) the sensitivity of the STOP-BANG measured for an apnea/hypopnea index (AHI, events per hour of sleep) >5, >15, and >30 was 82, 93, and 97% respectively. Corresponding negative predictive values were 44, 87, and 96%. Specificities were comparatively low (48, 40, and 33%). The STOP-BANG measured and STOP-BANG self-reported scores showed essentially equivalent test characteristics against polysomnography. Conclusions The STOP-BANG appears to have limited utility in a referred, sleep laboratory setting. Negative results help to identify some individuals as unlikely to have moderate-to-severe apnea, and may thereby prove useful in identification of patients who would benefit more from laboratory studies than home studies. A STOP-BANG in which all information is self-reported may be as effective as the original version, and has potential to facilitate research or community screening where good negative predictive value is required for an effective screening tool.