Abstract A major challenge in cancer treatment is the heterogeneous nature of the disease. This is particularly evident in breast cancer where gene expression profiling of whole tumours has identified multiple intrinsic subtypes of breast cancer. These subtypes are associated with differential prognoses and are correlated with previously identified clinical biomarkers (i.e., ER and HER2 status) used to stratify patients for targeted therapy. Despite recent studies demonstrating that elements within the tumour microenvironment can affect breast cancer progression and outcome, and that information contained within this compartment carries significant prognostic information for patient stratification above and beyond the information supplied by the intrinsic subtypes and existing therapeutic biomarkers, a limited understanding of stromal heterogeneity across the population has hindered the development of effective prognostic tools and targeted therapies directed against these processes. Here we perform expression profiling of the microdissected tumour-associated stromal components of 49 human breast tumours, and demonstrate that stromal heterogeneity can be captured by categorization into six classes which bear distinct molecular phenotypes. These stromal classes exhibit distinct biological functions and carry prognostic information independent of existing tumor-intrinsic biomarkers and molecular breast cancer subtypes. Specific combinations of stromal class and tumour subtype are significantly over- and under-represented; furthermore, simultaneous stratification of tumors in external datasets by both tumour subtype and stroma class identifies good- and poor-outcome cohorts within four of the five molecular breast cancer subtypes. The stroma classes identified here form the basis for an improved breast cancer classification scheme which takes the contribution of the microenvironment into account. Citation Format: Crista Thompson, Nicholas Bertos, Tina Gruosso, Greg Finak, Robert Lesurf, Sadiq M Saleh, Hong Zhao, Margarita Souleimanova, Sarkis Meterissian, Atilla Omeroglu, Michael T Hallett, Morag Park. A new breast cancer classification scheme based on novel classes of tumor stroma [abstract]. In: Proceedings of the Thirty-Seventh Annual CTRC-AACR San Antonio Breast Cancer Symposium: 2014 Dec 9-13; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2015;75(9 Suppl):Abstract nr P4-04-01.