Background There is ongoing debate about whether fusion or arthroplasty is superior in the treatment of single level cervical degenerative disc disease. Mainly because the intended advantage of arthroplasty over fusion, that is, the prevention of symptoms due to adjacent segment degeneration in the long term, is not confirmed yet. Until sufficient long‐term results become available, it is important to know whether results of one of the two treatments are superior to the other in the first one to two years. Objectives To assess the effects of arthroplasty versus fusion for radiculopathy or myelopathy, or both due to single level cervical degenerative disc disease. Search methods We searched the following databases for randomised controlled trials (RCTs): CENTRAL (The Cochrane Library 2011, Issue 2), MEDLINE, EMBASE, and EBMR. Additionally, we searched the System for Information on Grey Literature (SIGLE), subheading Biological and Medical Sciences, the US Food and Drug Administration (FDA) database on medical devices, and Clinicaltrials.gov to identify trials in progress. We also screened the reference list of all selected papers. Date of search: 25 May 2011. Selection criteria We included RCTs that directly compared any type of cervical fusion with any type of arthroplasty, with at least one year of follow‐up. Primary outcomes were arm pain, neck pain, neck‐related functional status, patient satisfaction, neurological outcome, and global health status. Secondary outcomes were the presence of (radiological) fusion, revision surgery at the treated level, secondary surgery on adjacent levels, segmental mobility of treated and adjacent levels, and work status. Data collection and analysis Study selection was performed independently by three review authors, and 'Risk of bias' assessment and data extraction were performed by two review authors. In case of missing data or insufficient information for a judgement about risk of bias, we tried to contact the study authors or the study sponsor. The data were entered into RevMan by one review author and subsequently checked by a second review author. We assessed the quality of evidence using GRADE. We analysed heterogeneity and performed sensitivity analyses for the pooled analyses. Main results We included nine studies (2400 participants), five of which had a low risk of bias. Eight of these studies were industry sponsored. The most important results showed low‐quality evidence for a small but significant difference in alleviation of arm pain at one to two years in favour of arthroplasty (mean difference (MD) ‐1.54; 95% confidence interval (CI) ‐2.86 to ‐0.22; 100‐point scale). A small study effect could not be ruled out for this outcome in the sensitivity analyses. This means that smaller studies (or small published subsets of larger studies) showed larger differences for this outcome, which may indicate publication bias. Also, moderate‐quality evidence showed a small difference in neck‐related functional status at one to two years in favour of arthroplasty (MD ‐2.79; 95% CI ‐4.73 to ‐0.85; 100‐point scale) and a small difference in neurological outcome in favour of arthroplasty (risk ratio (RR) 1.05; 95% CI 1.01 to 1.09). These two outcomes were robust to sensitivity analyses. For none of the primary outcomes, was a clinically relevant difference shown. Additionally, there was high‐quality evidence for a large, statistically significant difference in segmental mobility at one to two years (measured as degrees segmental range of motion) at the treated level (MD 6.90; 95% CI 5.45 to 8.35). There was low‐quality evidence that there was no statistically significant difference in secondary surgery at the adjacent levels at one to two years (RR 0.60; 95% CI 0.35 to 1.02). The latter was not robust to sensitivity analyses. Authors' conclusions There was a tendency for clinical results to be in favour of arthroplasty; often these were statistically significant. However, differences in effect size were invariably small and not clinically relevant for all primary outcomes. Significance was often gained or lost in the varying sensitivity analyses, probably owing to the relatively small number of studies, in combination with the small differences that were found. Given the fact that all of the included studies were not blinded, this could be due to patient or carer expectations. However, at this time both treatments can be seen as valid options with respect to results at a maximum of one to two years. Given the current absence of truly long‐term results, use of these mobile disc prostheses should still be limited to clinical trials. There was high‐quality evidence that the goal of preservation of segmental mobility in arthroplasty was met. A statistically significant effect on the incidence of secondary symptoms at adjacent levels, the primary goal of arthroplasty over fusion, was not found at one to two years. If there was a protective effect, this should become clearer over time. A future update, when studies with 'truly long‐term' results (five years or more) become available, should focus on this issue.