Thirty cancer patients with intractable pain were examined before and after five days of epidural opiate administration. The pain measurement was carried out with the German version of the McGill Pain Questionnaire. The patients with bone pain - the largest group - responded very well to the therapy. The spinal analgesia was less effective in patients whose complaints were mainly caused by visceral tumour growth or by the distension of pain-sensitive structures. The poorest results were seen in patients with predominantly nerve lesions and without bone involvement. However, a significant pain relief could be achieved in all patients due to the usually widespread sources of cancer pain.