Flumazenil is used for the reversal of benzodiazepine overdose. Serious adverse events (SAEs) including seizures and cardiac arrhythmias have been reported in patients treated with flumazenil, and the clinical advantage of flumazenil treatment has been questioned. The objective was to assess the risk of (S)AEs associated with the use of flumazenil in patients with impaired consciousness due to known or suspected benzodiazepine overdose. Studies included in the meta-analyses were identified by literature search in Medline, Cochrane Library and Embase using combinations of the words flumazenil, benzodiazepines, anti-anxiety agents, poisoning, toxicity and overdose. Randomised clinical trials (RCTs) in verified or suspected benzodiazepine overdose patients comparing treatment with flumazenil versus placebo were included. Pre-defined outcome measures were AEs, SAEs and mortality. Thirteen trials with a total of 994 randomised (990 evaluable) patients were included. AEs were significantly more common in the flumazenil group (138/498) compared with the placebo group (47/492) (risk ratio: 2.85; 95% confidence interval: 2.11-3.84; p < 0.00001). SAEs were also significantly more common in the flumazenil group compared with the placebo group (12/498 versus 2/492; risk ratio: 3.81; 95% CI: 1.28-11.39; p = 0.02). The most common AEs in the flumazenil group were agitation and gastrointestinal symptoms, whereas the most common SAEs were supraventricular arrhythmia and convulsions. No patients died during the blinded phase of the RCTs. The use of flumazenil in a population admitted at the emergency department with known or suspected benzodiazepine intoxication is associated with a significantly increased risk of (S)AEs compared with placebo. Flumazenil should not be used routinely, and the harms and benefits should be considered carefully in every patient.