Major advances in the techniques of discography since 1968, in conjunction with major strides in the evaluation of pain in recent years, prompted a study in which Holt's work on the specificity of discography was replicated and extended. For the present study, seven patients who had low-back pain and ten volunteers who had been carefully screened, with a questionnaire and a physical examination, to ensure that they had no history of problems with the back, had an injection at three levels, and all sessions were videotaped. After each injection, the participant was interviewed about the pattern and intensity of the pain, and then the discs were imaged with computed tomography. Five raters, who were blind to the condition of the participant, graded each disc as normal or abnormal on the basis of findings on magnetic resonance images that had been made before the injection and computed tomography (discography) were done. There was only one disagreement between the ratings that were made on the basis of the magnetic resonance images and those that were made on the basis of the discograms. Each participant's pain-related response was evaluated independently by two raters who viewed the videotapes of the discography. Inter-rater reliability was 0.99, 0.93, and 0.88 for the evaluation of intensity of the pain, pain-related behavior, and similarity of the pain to pain that the subject had had before the injection. In the asymptomatic individuals, the discogram was interpreted as abnormal for 17 per cent (five) of the thirty discs and for five of the ten subjects.(ABSTRACT TRUNCATED AT 250 WORDS)