Video-assisted mediastinoscopy (VAM) and endobronchial ultrasound-guided transbronchial needle aspiration (EBUS-TBNA) are the two most commonly used invasive methods for mediastinal staging of lung cancer. The objective of this review is to assess and compare the overall diagnostic values of VAM and EBUS-TBNA. PubMed, Embase, Web of Science and the Cochrane Library were searched for studies that evaluated EBUS-TBNA or VAM accuracy. Quantitative meta-analysis was used to pool sensitivity and specificity, and study quality was evaluated. Meta-regression was applied to indirectly compare EBUS-TBNA and VAM after adjusting quality score, study design, and station number. A total of ten studies with 999 EBUS-TBNA patients and seven studies with 915 VAM patients were included. Since the pooled specificity was 100% for both modalities, sensitivity was mainly analyzed. The pooled sensitivities for EBUS-TBNA and VAM were 0.84 (95% CI 0.79-0.88) and 0.86 (95% CI 0.82-0.90), respectively. Subgroup analyses of quality score, study design, station number and rapid on-site cytologic evaluation showed no significant influence on the overall sensitivity of the two modalities. After adjusting quality score, study design, and station number, the pooled sensitivities of VAM and EBUS-TBNA were not significantly different. However, more procedural complications and fewer false negatives (FN) were found with VAM than EBUS-TBNA. VAM and EBUS exhibited equally high diagnostic accuracy for mediastinal staging of lung cancer. Due to lower morbidity with EBUS-TBNA and fewer FN with VAM, EBUS-TBNA should be performed first, followed by VAM in the case of a negative needle result.