Purpose Self-reported measures have been widely used to indicate the presence of possible and probable sleep bruxism (SB) in both research and clinical situations. However, few studies have attempted to assess the diagnostic validity of this approach. The aim of this study was to estimate the diagnostic validity of self-reported measures of SB using an ambulatory single-channel electromyographic (EMG) device. Methods A total of 115 participants were enrolled and examined by standardized Research Diagnostic Criteria for Temporomandibular Disorders (RDC/TMD) including two questions related to SB: self-reported SB and morning-jaw symptoms. An ambulatory single-channel EMG device (GrindCare3™, Medotech A/S) was used for measuring jaw–muscle EMG activity during sleep for seven consecutive nights. Cut-off values for different measures of EMG activity (average, maximum and minimum) and the coefficient of variation (CV) were selected to divide participants into two groups, with higher or lower EMG activity or CV values. The sensitivity and specificity for each question and combination of them were calculated. Results Self-reported SB had the highest sensitivity (compared with morning-jaw symptoms) for all measures of EMG activity and CV, although the values were low to modest (average: 76.0%, maximum: 76.9%, minimum: 77.3%, CV: 61.0%). The specificity was low for both the questions related to the different measures of EMG activity and CV (35.1–52.4%). Conclusions This study indicated that the diagnostic validity of self-reported measures of SB was low to modest using an ambulatory EMG device assessment as a reference. Using only self-reported measures for the assessment of SB may not have a high validity, which should be taken into consideration in the clinical evaluation of patients.