Forty-six consecutive patients with neck pain and arm radiculopathy were treated with anterior cervical discectomy and fusion. All patients had neurological symptoms corresponding to a herniated disc and/or spondylosis at one or two cervical levels, verified by magnetic resonance imaging. The patients were stabilized with an anterior graft and randomized to either fixation with a CSLP plate or no internal fixation. Preoperatively and 2 years postoperatively the patients filled in a questionnaire that included a modified Million Index, a modified Oswestry Index and the Zung Depression Scale. They were also asked to register their pain in the arm and in the neck on a vertical visual analogue scale (VAS). At the 2-year follow-up, an unbiased observer graded the patients' clinical outcome using Odom's criteria. A test-retest procedure was carried out to examine the questionnaire reproducibility. In the group that was operated at one level, there was no significant improvement in any of the scores. Nevertheless, 81% of the patients were satisfied with the outcome of the surgery. All scores improved in the group operated at two levels. The pain in the neck and arm, as measured on a VAS, decreased in both groups. The improvement in arm pain was significantly more pronounced in patients operated with a plate at two levels compared to those who were operated without a plate. At the 2-year follow-up, patients with an excellent or good result according to Odom's criteria had a lower Million Index (P < 0.0005), Oswestry Index (P < 0.0005), and Zung (P = 0.024) score, than the group classified as fair or poor. There was a significant correlation (P < 0.0001 for all scores) between the test and retest results. We conclude that the modified Million Index and Oswestry Index are clinically useful tools in the evaluation of outcome after degenerative cervical disc surgery. The clinical benefits of plate fixation were minimal. The outcome after surgery, measured with the Oswestry Index, Million Index and VAS for arm and neck pain, seems to correlate well with the classification of outcome by Odom.