Objectives: International health agencies have promoted nontargeted universal (opt-out) HIV screening tests in different settings, including emergency departments (EDs). We performed a systematic review and meta-analysis to assess the testing uptake of strategies (opt-in targeted, opt-in nontargeted and opt-out) to detect new cases of HIV infection in EDs. Methods: We searched the Pubmed and Embase databases, from 1984 to April 2015, for opt-in and opt-out HIV diagnostic strategies used in EDs. Randomized controlled or quasi experimental studies were included. We assessed the percentage of positive individuals tested for HIV infection in each programme (opt-in and opt-out strategies). The mean percentage was estimated by combining studies in a random-effect meta-analysis. The percentages of individuals tested in the programmes were compared in a random-effect meta-regression model. Data were analysed using stata version 12. Quality assessments were performed using the Newcastle−Ottawa Scale. Results: Of the 90 papers identified, 28 were eligible for inclusion. Eight trials used opt-out, 18 trials used opt-in, and two trials used both to detect new cases of HIV infection. The test was accepted and taken by 75 155 of 172 237 patients (44%) in the opt-out strategy, and 73 581 of 382 992 patients (19%) in the opt-in strategy. The prevalence of HIV infection detected by the opt-out strategy was 0.40% (373 cases), that detected by the opt-in nontargeted strategy was 0.52% (419 cases), and that detected by the opt-in targeted strategy was 1.06% (52 cases). Conclusions: In this meta-analysis, the testing uptake of the opt-out strategy was not different from that of the opt-in strategy to detect new cases of HIV infection in EDs.