Surface electromyography (EMG) has been used extensively to estimate muscular load in studies of work related musculoskeletal disorders, especially for the trapezius muscle. The occurrences of periods of EMG silence (gaps), the time below a predetermined threshold level (muscular rest) and various percentiles of the amplitude distribution (APDF) are commonly used summary measures. However, the effects of the criteria used to calculate these measures (e.g., gap duration, threshold level, normalisation method) on the sensitivity of these measures to accurately differentiate work loads is not well known. Bilateral trapezius EMG was recorded, for a full workday, for 58 subjects following both maximal (MVE) and submaximal (RVE) reference contractions. Gap frequency, muscular rest, and percentiles were derived for eight fundamental work tasks. The calculations were performed using different gap duration criteria, threshold levels and normalisation methods.A gap duration of less than 1/2 s, and threshold level approximately 0.3% MVE for gap frequency, and approximately 0.5% MVE for muscular rest, were the criteria that optimised sensitivity to task differences. Minimal sensitivity to tasks and a high sensitivity to individuals was obtained using gap frequency with a threshold level of approximately 1% MVE. Normalisation to RVE, rather than MVE, improved sensitivity to differences between tasks, and reduced undesirable variability. Muscular rest was more sensitive to task differences than APDF percentiles.