OBJECTIVES: Hip resurfacing (HR) was developed for younger, more active patients, as a surgical alternative to total hip arthroplasty (THA). The safety of metal-on-metal HR is controversial with concerns expressed over adverse events and early device failure. We conducted a systematic review comparing primary HR to conventional THA for patients with hip osteoarthritis (OA).  Outcomes of interest were adverse event rates, early failure (revision/reoperation within 5 years), and post-operative component alignment.  METHODS: Studies were identified through electronic databases, grey literature and reference lists of included studies.  Inclusion criteria were: English language studies published after 1996 reporting adverse events, complications, safety issues or revision rates with respect to adults with primary hip OA, who underwent either primary HR or THA. Outcomes of interest included: revision, reoperation, dislocation, infection/sepsis, femoral neck fractures, time to revision, rates of early failure, mortality, and post-operative component alignment.  Results were reported per 1000 person years for comparability and stratified by age, publication date and market status (in-use and discontinued). RESULTS:  A total of 7421 abstracts were identified and screened.  Of these 384 full text articles were reviewed, 236 of which were included in this analysis. For all devices, those in-use and discontinued, dislocations were more frequent in THA, while revisions and reoperations were more frequent in HR.  An analysis of only devices currently in-use determined that early revisions/reoperations were more frequent in HR than in THA (range 3.1-13.1 vs. 1.9-7.5 per 1000 person years, respectively). CONCLUSIONS: This systematic review provides clinicians and health policy makers with information on adverse events based on device type and market status. Rates of adverse events vary by device type, making it challenging to conclude which device is more effective. Findings highlight the importance of differentiating market status and standardizing exposure time by 1000 person years to facilitate comparisons between studies.