Myofeedback seems a promising technique to treat myalgia related to computer work. A garment with embedded electrodes allows independent use by the subject. However, this inevitably results in electrode dislocations that can cause such variability in the electromyogram (EMG) parameters that the treatment fails. The objective was to investigate the variability introduced by electrode displacement on EMG root mean square (RMS) and relative rest time (RRT) and the effect of a normalisation procedure using a reference contraction to compensate for this. Dislocation was simulated using simultaneous recordings of pairs of electrodes situated 15mm apart on the trapezius muscle, surrounding a central electrode pair at the standard position. To assess sensitivity for electrode displacement, intra-subject variability was used. To assess the effects of normalisation, both intra and inter-subject variability were used. Results indicated a large sensitivity of RMS to electrode displacements, with no differences between typing (42%) and the reference task (43%). No systematic changes were found relative to a specific electrode dislocation. Normalisation of RMS resulted in a decrease in the sensitivity for electrode displacement (33%), but did not decrease inter-subject variability (27%). The median value of the intra-subject variability of RRT, using a fixed threshold, was 4.9%, whereas the normalised threshold did not decrease the intra-subject variability (7%). The results suggest that RRT is suitable for individual follow-up measurements and applications such as myofeedback, without the need for a normalisation procedure. The use of RMS requires considerably more care, owing to its large sensitivity to electrode displacement and lack of substantial effects of normalisation.