Objectives: To investigate the factors affecting the topography of trapezius muscle EMG, multichannel recordings were made at different forces of isometric shoulder elevation and during fatiguing exercise.Methods: Twenty-eight channels of monopolar EMG were recorded from an array of 4x7 electrodes placed on the upper trapezius muscle. From the monopolar EMG and the bipolar derivations the root mean square (RMS m o n o p o l a r , RMS b i p o l a r ) and power spectrum median frequency (MF m o n o p o l a r , MF b i p o l a r ) were calculated.Results: The maximum RMS m o n o p o l a r was located above the middle part of the trapezius muscle, where a minimum was found for RMS b i p o l a r . The cranial-caudal RMS distribution shifted upwards when the force was increased from 50 to 100% MVC and during fatigue. MF b i p o l a r showed a peak above the endplate region, where the MF m o n o p o l a r was low. During fatigue the normalized MF slope was independent of the cranial-caudal electrode position, but MF m o n o p o l a r decreased most strongly at positions above the endplate region, where MF b i p o l a r decreased less.Conclusions: While the changes in MF reflected metabolic properties and volume conduction phenomena in the muscle, changes in RMS reflected a compensation for the fatigue processes within the muscle. The RMS changes in fatigue can be explained by the direction of the fibres involved in shoulder elevation.