Objective To systematically review the diagnostic value of transcranial doppler for ≥50% middle cerebral artery stenosis, and evaluate the best cutoff value of blood flow velocity. Methods All data were independently searched and extracted in duplicate by two investigators. By searching PubMed, Web of science, Cochrane liberary, CNKI, Wanfang and VIP database from 1 January 2000 to 30 September 2017, all of the eligible articles which conformed to the inclusion and exclusion criteria were harvested. The quality of each included literature was evaluated by QUADAS tools. Statistical analysis was conducted by the Meta-Disc software. Results A total of 16 articles were analyzed in this meta-analysis. The best cutoff values of the mean flow velocity (MFV) and the peak systolic velocity (PSV) were ≥100cm/s and ≥140cm/s. Using digital subtraction angiography (DSA) as gold standard, the pooled sensitivity, specificity, positive likelihood radio and negative likelihood ratio were 0.77, 0.94, 12.68 and 0.23 for MFV≥100cm/s, and 0.79, 0.96, 19.85 and 0.22 for PSV≥140cm/s. Using DSA, magnetic resonance angiography (MRA) or computed tomographic angiography (CTA) as gold standard, the pooled sensitivity, specificity, positive likelihood radio and negative likelihood ratio were 0.71, 0.94, 5.54 and 0.25 for MFV≥100cm/s, and 0.75, 0.91, 10.00 and 0.27 for PSV≥140cm/s. Conlusions When MFV≥100cm/s and PSV≥140cm/s, TCD shows high value in the diagnosis of ≥50% MCA stenosis, and it can be used as an effective screening test in clinical practice.