EBUS-TBNA is a recent mediastinal staging and diagnostic technique. We have previously reported superior characterisation with 21G biopsies over 22G biopsies for benign and malignant mediastinal nodes. A new 19G needle now exists but there are limited studies. We hypothesised 19G biopsies would improve both benign and malignant characterisation due to larger samples. We retrospectively analysed sequential patients referred for EBUS-TBNA with unexplained mediastinal adenopathy performed with 19G, 21G and 22G needles respectively (100 patients each). Contingency table analysis was performed. There were no complications. Sensitivity for malignancy was highest in the 19G group (95.7% versus 94.7% and 87.5%, respectively). The 19G group had higher mean lymph node size (19.4mm versus 18.6mm and 13.5mm, respectively), the highest proportion of lymphoma (9% versus 5% and 0%, respectively), the lowest proportion of NSCLC-NOS (2% versus 12% and 5%, respectively), the highest proportion of subcharacterised benign disease (89.6% versus 69.8% and 37.9%, respectively). This large single centre retrospective UK study suggests the 19G needle appears safe with the suggestion of better sensitivity for malignancy subcharacterisation of benign disease but this requires further study in adequately powered comparative controlled studies with univariate and multivariate analysis.