Clinical emergencies can be defined as unpredictable events that necessitate immediate intervention. Safety critical industries have acknowledged the difficulties of responding to such crises. Strategies to improve human performance and mitigate its limitations include the provision and use of cognitive aids, a family of tools that includes algorithms, checklists and decision aids. This systematic review evaluates the usefulness of cognitive aids in clinical emergencies. Following a systematic search of the electronic databases, we included 13 randomised controlled trials, reported in 16 publications. Each compared cognitive aids with usual care in the context of an anaesthetic, medical, surgical or trauma emergency involving adults. Most trials used only clinicians in the development and testing of the cognitive aids, and only some trials provided familiarisation with the cognitive aids before they were deployed. The primary outcome was the completeness of care delivered to the patient. Cognitive aids were associated with a reduction in the incidence of missed care steps from 43.3% to 11% (RR (95%CI) 0.29 (0.15-0.16); p < 0.001), and the quality of evidence was rated as moderate. The use of cognitive aids was related to decreases in the incidence of errors, increases in the rate of correctly performed steps and improvement in the clinical teamwork skills scores, non-technical skills scores, subjective conflict resolution scores and the global assessment of team performance. Cognitive aids had an inconsistent influence on the time to first intervention and time to complete care of the patient's condition. It is possible that this was a reflection of how common or rare the crisis in question was as well as the experience and expertise of the clinicians and team. Sufficient thought should be applied to the development of the content and design of cognitive aids, with consideration of the pre-existing guideline ecosystem. Cognitive aids should be tested before their deployment with adequate clinician and team training.