Nefopam, a centrally acting analgesic, has been used in the surgical setting in many countries since the mid-1970s. However, clinical trials provide contflicting results for its analgesic potency. We performed a systematic search (multiple databases, bibliographies, any language, to January 2008) for randomized, placebo-controlled trials of nefopam for the prevention of postoperative pain. Data were combined using classic methods of meta-analyses and were expressed as weighted mean difference (WMD), relative risk (RR), and number needed to treat/harm (NNT/H) with 95% confidence interval (CI). Nine trials (847 adult patients, 359 received nefopam) were included. Nefopam (cumulative doses, 20-160 mg) was given orally or i.v., as single or multiple doses, or as a continuous infusion. Compared with placebo, cumulative 24 h morphine consumption was decreased with nefopam: WMD -13 mg (95% CI -17.9 to -8.15). Pain intensity at 24 h was also decreased: on a 100 mm visual analogue scale, WMD -11.5 mm (95% CI -15.1 to -7.85). The incidence of tachycardia was increased with nefopam (RR 3.12, 95% CI 1.11-8.79; NNH 7), as was the incidence of sweating (RR 4.92, 95% CI 2.0-12.1; NNH 13). There is limited evidence from the published literature that nefopam may be a useful non-opioid analgesic in surgical patients. The analgesic potency seems to be similar to non-steroidal anti-inflammatory drugs. However, dose responsiveness and adverse effect profile remain unclear, and the role of nefopam as part of multimodal analgesia needs to be established. Data in children are lacking.