Randomised controlled trials (RCTs) of cervical disc arthroplasty vs fusion generally show slightly more favourable results for arthroplasty. However, RCTs in surgery often have limited external validity, since they involve a select group of patients who fit very rigid admission criteria and who are prepared to subject themselves to randomisation. The aim of this study was to examine whether the findings of RCTs are verified by observational data recorded in our Spine Center in association with the Spine Society of Europe Spine Tango surgical registry. Patients undergoing fusion/stabilisation or disc arthroplasty for degenerative cervical spinal disease were selected for inclusion. They completed a questionnaire pre-operatively and at 12 and 24 months follow-up (FU). The questionnaire comprised the multidimensional Core Outcome Measures Index (COMI; 0-10 scale) and, at FU, questions on global outcome and satisfaction with treatment (5-point scales, dichotomised to "good" and "poor"), re-operation and patient-rated complications. The surgeon completed a Spine Tango Surgery form. The outcome data from 266 (208 fusion, 58 arthroplasty) out of 284 eligible patients who had reached 12 months FU, and 169 (139 fusion, 30 arthroplasty) out of 178 who had reached 24 months FU, were included. Patients with cervical disc arthroplasty were younger [46 (SD 8) years vs 56 (SD 11) years for fusion; P < 0.05], had less comorbidity (P < 0.05), more often had only mono-segmental pathology (69% arthroplasty, 47% fusion) and only one type of degenerative pathology (69% arthroplasty, 46% fusion). Surgical complication rates were similar in each group (arthroplasty, 1.5%; fusion, 2.6%). The reduction in the COMI score was significantly greater in the arthroplasty group (at 12 months, 4.8 (SD 3.0) vs 3.7 (SD 2.9) points for fusion, and at 24 months 5.1 (SD 2.8) vs 3.8 (SD 2.9) points; each P < 0.05). In the arthroplasty group, a "good" global outcome was recorded in 90% patients (at 12 months) and 93% (at 24 months); in the fusion group the figures were 80 and 82%, respectively (group differences at each timepoint, P > 0.09). Satisfaction with treatment was similar in both groups (89-93%), at each timepoint. In multiple regression analysis, treatment group was of borderline significance as a unique predictor of the change in COMI at FU (P = 0.059 at 12 months, P = 0.055 at 24 months) in a model in which known confounders (age, comorbidity, number of affected levels) were controlled for. Being in the arthroplasty group was associated with an approximately 1-point greater reduction in the COMI score at FU. The results of this observational study appear to support those of the RCTs and suggest that, in patients with degenerative pathology of the cervical spine, disc arthroplasty is associated with a slightly better outcome than fusion. However, given the small size of the difference, its clinical relevance is questionable, especially in view of the a priori more favourable outcome expected in the arthroplasty group due to the more rigorous selection of patients.