Perineural dexamethasone has gained popularity in regional anaesthesia to prolong the duration of analgesia, but its advantage over systemic administration is disputed. The objective of this meta-analysis was to compare the analgesic efficacy of both routes of administration during peripheral nerve block. The methodology followed the PRISMA statement guidelines. The primary outcome was the duration of analgesia analysed according to the type of local anaesthetic administered (bupivacaine or ropivacaine). Secondary outcomes included cumulative opioid consumption in morphine i.v. equivalents, pain scores, and complication rates (neurological complications, infection, or hyperglycaemia). Eleven controlled trials, including 914 patients, were identified. The duration of analgesia was significantly increased with perineural dexamethasone vs systemic dexamethasone by a mean difference of 3 h [95% confidence interval (CI): 1.4, 4.5 h; P=0.0001]. Subgroup analysis revealed that the duration of analgesia was increased by 21% with bupivacaine (mean difference: 4.0 h; 95% CI: 2.8, 5.2 h; P<0.00001) and 12% with ropivacaine (mean difference: 2.0 h; 95% CI: -0.5, 4.5 h; P=0.11). The quality of evidence for our primary outcome was moderate according to the GRADE system. There were no significant differences in other secondary outcomes. No neurological complications or infections were reported. Glucose concentrations were not increased when dexamethasone was injected systemically, but this outcome was reported by only two trials. There is, therefore, moderate evidence that perineural dexamethasone combined with bupivacaine, but not ropivacaine, slightly prolongs the duration of analgesia, without an impact on other pain-related outcomes, when compared with systemic dexamethasone. Injection of perineural dexamethasone should be cautiously balanced in light of the off-label indication for this route of administration.