In a previous study on a small series of breast cancers, we developed objective methods for the assessment of vascular invasion (VI), using immunohistochemical staining. We found that VI was predominantly lymphovascular invasion (LVI), with minimal contribution of blood vascular invasion (BVI). The aims of the current study were: (a) to assess the frequency, extent and prognostic role of LVI and BVI in a large, well-characterized series of LN-negative breast cancers; and (b) to assess the ability of VI to stratify early breast cancer into different prognostic groups. Paraffin-embedded sections from 1005 lymph-node (LN)-negative primary invasive breast cancers were stained for CD34, CD31 and podoplanin/D240 to detect BVI and LVI. VI lesions were assessed and the results were correlated with clinicopathological criteria and survival. VI was detected in 218 (22%); 211/218 (97%) were LVI, while BVI was detected in 7/218 (3%). The frequency of LVIs/section ranged from 1 to 79, with no significant difference between the frequency of LVI and outcome. The presence of LVI was significantly associated with adverse disease-free interval (DFI) and poor overall survival (OS) in both univariate and multivariate analyses. The results from the study indicated that VI in early stage breast cancer is predominantly LVI and that its objective assessment is a powerful independent prognostic factor. Efforts to detect early metastatic activity, such as diligent pathological examination of sentinel LN biopsies would be complimented by the objective evaluation of VI status of the primary tumour. VI status should be included routinely in breast cancer staging systems. Copyright © 2010 Pathological Society of Great Britain and Ireland. Published by John Wiley & Sons, Ltd.