Right hemispheric brain damage is often accompanied by visual neglect of contralesional hemispace as well as profound visuospatial and visuoconstructive disorders. Despite some recent improvements in the treatment of neglect, few therapeutic approaches exist for disorders of visuospatial cognition and none has dealt with the question of how both disorders can be treated in combination. In the present cross-over rehabilitation study, thirteen neglect patients first received visual exploration training for six weeks followed by visuospatial training for four weeks. Eleven patients had a right-hemispheric lesion, 2 a left-hemispheric lesion. Ten subjects had parietal cortical lesions, 3 lesions were outside the parietal lobe. In the first treatment phase the patients received training, using computer controlled devices, intended to improve visual search and reading strategies. In the second phase of treatment, visual orientation discrimination was trained on a PC-monitor with verbal feedback by the therapist. Significant improvements were obtained after exploration training regarding the extent of visual search-field witbin the neglected hemifield, visual search performance in the whole visual field, reading time and reading errors but there was no change after visuospatial training for any of these measures. The reverse pattern was observed for visuospatial training: significant improvements were observed in line orientation judgements, clock perception, mental transposition of angles, horizontal writing and visuoconstructive abilities, but there was no change in any of these tests during exploration training. Feedback-based training of visual orientation discrimination revealed rapid (within 8-15 sessions) and stable improvements in this ability and a transfer to spatially related tasks.No significant correlations were obtained between training-related improvements and the time since brain damage for 9 of 10 tests, indicating that spontaneous recovery had no effect on the improvements during the two treatments. Likewise, no influence of age and gender on treatment progress was found. The phase-specific improvements in this study suggest that specific and differential treatment is necessary and effective for the treatment of visual exploration and visuospatial disorders in patients with visual neglect. Moreover, they indicate that feedback-based visual learning may prove an effective device in the neurorehabilitation of brain-damaged patients.