The aim of the study was to evaluate four different modes of human-computer interaction. The modes were: use of the keyboard alone as input device, use of keyboard and mouse, use of keyboard and mouse with a three-dimensional arm support, and use of a keyboard with a Trackpoint device in its centre. Ten women and 10 men volunteered to participate. Questions asked were whether working in the different modes influenced shoulder and forearm muscular load differently during word processing, and how much strain on the neck, shoulder and arms subjects perceived in the different modes. Muscular load was studied with electromyography in three shoulder muscles and three forearm muscles. The subjects also rated the different modes in one questionnaire concerning perceived strain and in one concerning preference for any of the modes tested. Intra-individual analysis for each muscle and mode showed two possible ways of decreasing the strain from computer mouse work on the shoulder muscles--either to use Trackpoint or to use the mouse combined with the movable arm support. However, both of these computer-interaction modes increased the muscular load in the hand and forearm.