The heterogeneity of endothelial cells (ECs), lining blood vessels, across tissues remains incompletely inventoried. We constructed an atlas of >32,000 single-EC transcriptomic data from 11 tissues of the model organism Mus musculus. We propose a new classification of EC phenotypes based on transcriptome signatures and inferred putative biological features. We identified top-ranking markers for ECs from each tissue. ECs from different vascular beds (arteries, capillaries, veins, lymphatics) resembled each other across tissues, but only arterial, venous and lymphatic (not capillary) ECs shared markers, illustrating a greater heterogeneity of capillary ECs. We identified high-endothelial-venule and lacteal-like ECs in the intestines, and angiogenic ECs in healthy tissues. Metabolic transcriptomes of ECs differed amongst spleen, lung, liver, brain and testis, while being similar for kidney, heart, muscle and intestines. Within tissues, metabolic gene expression was heterogeneous amongst ECs from different vascular beds, altogether highlighting large EC heterogeneity.