Background The role of low-level laser therapy (LLLT) in the management of carpal tunnel syndrome (CTS) is controversial. While some trials have shown distinct advantages of LLLT over placebo and some other non-surgical treatments, other trials have not. Objectives To assess the benefits and harms of LLLT versus placebo and versus other non-surgical interventions in the management of CTS. Search methods On 9 December 2016 we searched CENTRAL, MEDLINE, Embase, and Science Citation Index Expanded for randomised controlled trials (RCTs). We also searched clinical trial registries for ongoing studies. We checked the references of primary studies and review articles, and contacted trial authors for additional studies. Selection criteria We considered for inclusion RCTs (irrespective of blinding, publication status or language) comparing LLLT versus placebo or non-surgical treatment for the management of CTS. Data collection and analysis Two review authors independently identified trials for inclusion and extracted the data. For continuous outcomes, we calculated the mean difference (MD) or standardised mean difference (SMD) with a 95% confidence interval (CI) using the random-effects model, calculated using Review Manager. For dichotomous data, we reported risk ratio (RR) and 95% CI. Main results We identified 22 trials randomising 1153 participants that were eligible for inclusion; nine trials (525 participants, 256 randomised to LLLT) compared LLLT with placebo, two (150 participants, 75 randomised to LLLT) compared LLLT with ultrasound, one compared LLLT with placebo and LLLT with ultrasound, two compared LLLT with steroid injection, and one trial each compared LLLT with other non-surgical interventions: fascial manipulation, application of a pulsed magnetic field, transcutaneous electrical nerve stimulation (TENS), steroid injection, tendon gliding exercises, and applying a wrist splint combined with non-steroidal anti-inflammatory drugs. Three studies compared LLLT as part of multiple interventions. Risk of bias varied across the studies, but was high or unclear in most assessed domains in most studies. Most studies were small, with few events, and effect estimates were generally imprecise and inconsistent; the combination of these factors led us to categorise the quality of evidence for most outcomes as very low or, for a small number, low. At short-term follow-up (less than three months), there was very low-quality evidence for any effect over placebo of LLLT on CTS for the primary outcome of Symptom Severity Score (scale 1 to 5, higher score represents worsening; MD -0.36, 95% CI -0.78 to 0.06) or Functional Status Scale (scale 1 to 5, higher score represents worsened disability; MD -0.56, 95% CI -1.03 to -0.09). At short-term (less than three months) follow-up, we are uncertain whether LLLT results in a greater improvement than placebo in visual analogue score (VAS) pain (scale 0 to 10, higher score represents worsening; MD -1.47, 95% CI -2.36 to -0.58) and several aspects of nerve conduction studies (motor nerve latency: higher score represents worsening; MD -0.09 ms, 95% CI -0.16 to -0.03; range 3.1 ms to 4.99 ms; sensory nerve latency: MD -0.10 ms, 95% CI -0.15 to -0.06; range 1.8 ms to 3.9 ms), as the quality of the evidence was very low. When compared with placebo at short-term follow-up, LLLT may slightly improve grip strength (MD 2.58 kg, 95% CI 1.22 to 3.95; range 14.2 kg to 25.23 kg) and finger-pinch strength (MD 0.94 kg, 95% CI 0.43 to 1.44; range 4.35 kg to 5.7 kg); however, the quality of evidence was low. Only VAS pain and finger-pinch strength results reached the minimal clinically important difference (MCID) as previously published. We are uncertain about the effect of LLLT in comparison to ultrasound at short-term follow-up for improvement in VAS pain (MD 2.81, 95% CI 1.21 to 4.40) and motor nerve latency (MD 0.61 ms, 95% CI 0.27 to 0.95), as the quality of evidence was very low. When compared with ultrasound at short-term follow-up, LLLT may result in slightly less improvement in finger-pinch strength (MD -0.71 kg, 95% CI -0.94 to -0.49) and motor nerve amplitude (MD -1.90 mV, 95% CI -3.63 to -0.18; range 7.10 mV to 9.70 mV); however, the quality of evidence was low. There was insufficient evidence to assess the long-term benefits of LLLT versus placebo or ultrasound. There was insufficient evidence to show whether LLLT is better or worse in the management of CTS than other non-surgical interventions. For all outcomes reported within these other comparisons, the quality of evidence was very low. There was insufficient evidence to assess adverse events, as only one study reported this outcome. Authors' conclusions The evidence is of very low quality and we found no data to support any clinical effect of LLLT in treating CTS. Only VAS pain and finger-pinch strength met previously published MCIDs but these are likely to be overestimates of effect given the small studies and significant risk of bias. There is low or very low-quality evidence to suggest that LLLT is less effective than ultrasound in the management of CTS based on short-term, clinically significant improvements in pain and finger-pinch strength. There is insufficient evidence to support LLLT being better or worse than any other type of non-surgical treatment in the management of CTS. Any further research of LLLT should be definitive, blinded, and of high quality.