A novel integrative model for simulating the surface EMG signal during pathological tremor is presented. The model combines neuromuscular elements, biomechanical descriptions, and surface EMG generation. First single motor unit spike trains are generated based on the sum of the simulated descending drive, afferent input and an oscillatory noise causing tremor. Based on this activity pattern, the muscle force is estimated, from which the limb movement is derived. The surface EMG is simulated as the sum of the surface action potentials generated by the active motor units. The model was able to simulate several features of tremor that have been previously observed experimentally, including the spectral characteristics of the surface EMG during tremor and the pattern of activity of single motor units.