We aimed to assess effectiveness of simulation for teaching medical students critical care medicine and to assess which simulation methods were most useful. We searched AMED, EMBASE, MEDLINE, Education Resources Information Centre, British Education Index, Australian Education Index, and bibliographies and citations, in July 2013. Randomized controlled trials comparing effectiveness of simulation with another educational intervention, or no teaching, for teaching medical students critical care medicine were included. Assessments for inclusion, quality, and data extraction were duplicated and results were synthesized using meta-analysis.We included 22 randomized control trials (n = 1325). Fifteen studies comparing simulation with other teaching found simulation to be more effective [standardized mean difference (SMD) = 0.84; 95% confidence interval (CI) = 0.43 to 1.24; P < 0.001; I = 89%]. High-fidelity simulation was more effective than low-fidelity simulation, and subgrouping supported high-fidelity simulation being more effective than other methods. Simulation improved skill acquisition (SMD = 1.01; 95% CI = 0.49 to 1.53) but was no better than other teaching in knowledge acquisition (SMD = 0.41; 95% CI = -0.09 to 0.91).