The successful reduction of postoperative discomfort is of great significance. This review aims to evaluate the efficacy of low-level laser therapy (LLLT) for the reduction of complication caused by impacted mandibular third molars extraction. An extensive literature search up to October 2013 for randomized controlled trials (RCTs) was performed through CENTRAL, PubMed, Embase, Medline, and CNKI. Six RCTs in which involves 193 participants are included in the meta-analysis. Among them, three RCTs exhibit a moderate risk of bias, while the other three show a high bias risk. Compared with placebo laser/control group, pain is significantly reduced with LLLT on the first day (mean difference [MD] = -2.63, 95% confidence interval [CI] -4.46 to -0.79, P = 0.005). The superiority of LLLT in pain control persists on the second day (MD = -2.34, 95% CI -4.61 to -0.06, P = 0.04) and the third day (MD = -3.40, 95% CI -4.12 to -2.68, P < 0.00001). Moreover, LLLT reduces an average of 4.94 mm (MD = 4.94, 95 % CI 1.53 to 8.34, P = 0.004) of trismus compared with placebo laser irradiation in the first 3 days. On the seventh day, the superiority of LLLT also persists (MD = 3.24, 95% CI 0.37 to 6.12, P = 0.03). In the first 3 days after surgery, extraoral irradiation (MD = -0.69, 95% CI -1.30 to -0.08, P = 0.03) and intraoral combined with extraoral irradiation (MD = -0.65, 95% CI -1.15 to -0.15, P = 0.01) reduced facial swelling significantly. On the seventh day, the intraoral combined with extraoral irradiation group (MD = -0.32, 95% CI -0.59 to -0.06, P = 0.02) still showed benefit in relieving facial swelling. However, because of the heterogeneity of intervention and outcomes assessment and risk of bias of included trials, the efficacy is proved with limited evidence. In the future, well-designed RCTs with larger sample size will be required to provide clearer recommendations.