Measurement of physical exposure to musculoskeletal disorder risk factors must generally be performed directly in the field to assess the effectiveness of ergonomic interventions. To perform such an evaluation, the reliability of physical exposure measures under similar field conditions must be known. The objectives of this study were to estimate the reliability of physical exposure measures performed in the field and to establish the baseline values of physical exposure in sign language interpreters (SLI) before the implementation of an intervention. The electromyography (EMG) of the trapezius muscles as well as the wrist motions of the dominant arm were measured using goniometry on nine SLI on four different days. Several exposure parameters, proposed in the literature, were computed and the generalizability theory was used as a framework to assess reliability. Overall, SLI showed a relatively low level of trapezius muscle activity, but with little time at rest, and highly dynamic wrist motions. Electromyography exposure parameters showed poor to moderate reliability, while goniometry parameter reliability was moderate to excellent. For EMG parameters, performing repeated measurements on different days was more effective in increasing reliability than extending the duration of the measurement over one day. For goniometry, repeating measurements on different days was also effective in improving reliability, although good reliability could be obtained with a single sufficiently long measurement period.