Pain is now regarded as 'the fifth vital sign' and patients are frequently asked to score the intensity of their pain on a numerical pain rating scale (NPRS). However, the use of a unidimensional scale is questionable in view of the belief, overwhelmingly supported by clinical experience as well as by empirical evidence from multidimensional scaling and other sources, that pain has at least two dimensions: somatosensory qualities and affect. We used a Chinese translation of the 101 descriptor multidimensional affect and pain survey (MAPS) questionnaire to determine the relative contributions of various dimensions of postoperative pain to a patient's score on a unidimensional NPRS. MAPS and NPRS were administered postoperatively to 69 patients with descending colon carcinoma who were recovering from left hemi-colectomy. Multiple linear regression revealed that the emotional pain qualities supercluster (P=0.0005) and four of its eight subclusters, anxiety, depressed mood, fear and anger, significantly (P=0.001-0.007) predicted a patient's score on the unidimensional NPRS. Notably, none of the 17 subclusters in the somatosensory pain qualities supercluster predicted NPRS scores. It may be concluded that patient scores on unidimensional pain intensity scales reflect the emotional qualities of pain much more than its sensory intensity or other qualities. Accordingly such scales are poor indicators of analgesic requirement. The results also suggest that patients' postoperative anxiety and depression are inadequately treated. Based on our findings we present six unidimensional scales that should yield a more accurate assessment of the sources of a patient's pain.