The optimal approach to operative treatment of humeral shaft fractures remains debatable. Previously published trials have been limited in size and have been inconclusive regarding important patient outcome variables following treatment with either intramedullary nails or plates. We conducted a meta-analysis of available trials comparing treatment of humeral shaft fractures.We performed a literature search from 1967 to November 2007 in the main medical search engines and selected 4 randomized trials that compared nails and plates in patients with humeral shaft fractures and that reported on complications due to surgery. We statistically pooled patient data using standard meta-analytic approaches. Our primary outcome was the total complication rate, comprised of all complications listed in the articles included. Secondary outcomes included non-union, infection, nerve palsy, and reoperation rate. Methodology was assessed using the CLEAR NPT.When pooling the data of the 4 trials (n = 203 patients), we did not find a statistically significant difference between implants in the rate of total complications, non-union, infection, nerve-palsy, or the need for reoperation. The studies included were small and had methodological limitations.Our meta-analysis suggests stastistically insignificant differences between plates and nails in the treatment of humeral shaft fractures. Small sample sizes, study heterogeneity, and methodological limitations argue strongly for a definitive, large trial. We recommend that this trial should be a randomized controlled trial with appropriate allocation of patients and blinding of patients and care providers and outcome assessors, and that it should include patient-important outcomes.