The purpose of this experimental crossover study was to evaluate to what extent the therapeutic use of music would decrease pain in patients with cancer who were receiving scheduled analgesics. Baseline data were collected for three days. Subjects then were assigned randomly to listen to their preference of seven types of relaxing music or a control (a 60-cycle hum) twice daily for three days. Then they crossed over into the alternate group for the next three days. Finally, each subject returned to a follow-up baseline period. Pain, the dependent variable, and mood, which was proposed as an intervening variable, were measured by visual analogue scales. The convenience sample included 15 outpatients with cancer, 12 female and 3 male, ages 20 through 87. Results of the McGill Pain Questionnaire (MPQ), a reliable and valid multidimensional instrument administered upon entry into the study, indicated that the study sample was comparable to other samples of patients with cancer who were in pain. There was an inconsistent relation between pain and mood. The effect of the music on pain varied by individual; 75% had at least some response and 47% had a moderate or great response. Multivariate Analysis of Variance (MANOVA) indicated a statistically significant decrease in pain from using either the music or sound, but there was no effect on mood. Although the mean percentage of change in pain for music was twice that for sound, the results did not differ statistically. The findings support the use of music as an independent nursing intervention to relieve pain.