Acupuncture is frequently used for pain relief in labour, but the evidence is not clear. To critically evaluate the evidence for or against acupuncture for labour pain management. Nineteen electronic databases, including English, Korean, Japanese, and Chinese databases, were systematically searched. All randomised controlled trials (RCTs) involving women receiving acupuncture alone, or as an adjunct to conventional analgesia, for pain relief in labour were considered. Pain intensity on a 100-mm visual analogue scale (VAS; 0, no pain; 100, worst pain) and use of other analgesic methods were used as primary outcomes, and for statistical pooling. Maternal/fetal outcomes were secondary outcomes, and adverse events were also recorded. Risk of bias was assessed regarding randomisation, allocation concealment, blinding, incomplete outcome data, selective outcome reporting, and other biases. Ten RCTs involving 2038 women were included. VAS for pain intensity data were available in seven studies; the meta-analysis shows that acupuncture was not superior to minimal acupuncture at 1 hour (pooled mean difference -8.02; 95% CI -21.88, 5.84; I(2) = 94%) and at 2 hours (-10.15; 95% CI -23.18, 2.87; I(2) = 92%). Patients reported significantly reduced pain by 4 and 6% during electroacupuncture (EA) treatment at 15 (-4.09; 95% CI -8.05, -0.12) and 30 minutes (-5.94; 95% CI -9.83, -2.06), compared with placebo EA, but the effect was not maintained afterwards. Compared with no intervention, acupuncture reduced pain by only 11% for the first 30 minutes (-10.56; 95% CI -16.08, -5.03). In trials where acupuncture was compared with conventional analgesia, women receiving acupuncture required less meperidine (pooled risk ratio 0.20; 95% CI 0.12, 0.33) and other analgesic methods (0.75; 95% CI 0.66, 0.85). No acupuncture-related adverse events were reported. Most trials did not blind participants, care providers and/or evaluators. The evidence from RCTs does not support the use of acupuncture for controlling labour pain. The primary studies are diverse and often flawed. Further research seems warranted.