AIM: To examine whether visual restitution training (VRT) is able to change absolute homonymous field defect, assessed with fundus controlled microperimetry, in patients with hemianopia. METHODS: 17 patients with stable homonymous visual field defects before and after a 6 month VRT period were investigated with a specialised microperimetric method using a scanning laser ophthalmoscope (SLO). Fixation was controlled by SLO fundus monitoring. The size of the field defect was quantified by calculating the ratio of the number of absolute defects and the number of test points; the training effect E was defined as the difference between these two ratios before and after training. A shift of the entire vertical visual field border by 1 degrees would result in an E value of 0.14. RESULTS: The mean training effect of all right eyes was E = 0.025 (SD 0.052) and all left eyes E = 0.008 (SD 0.034). In one eye, a slight non-homonymous improvement along the horizontal meridian occurred. CONCLUSIONS: In one patient, a slight improvement along the horizontal meridian was found in one eye. In none of the patients was an explicit homonymous change of the absolute field defect border observed after training.