Dengue virus (DENV) is the most important arthropod-borne virus globally, causing dengue fever and life-threatening severe dengue typified by vascular leakage and shock. Dengue is a multifactorial disease, but despite ultimately impacting the cardiovascular system, mechanistically dengue microvascular dysfunction remains poorly understood. Most studies are performed on capillary-derived endothelial cells and do not consider complex multi-cellular in vivo interactions, or vessel-specific and tissue-specific pathology. We developed complex models of dengue microvascular dysfunction by incorporating human perivascular cells (pericytes) and capillary or lymphatic endothelial cells into co-culture models, including microvascular spheroids, vessel-like networks grown in a 3D matrix, and mono- and multi-vessel organs-on-a-chip incorporating circulatory flow. We found that pericytes are crucial for amplifying leakage induced by the DENV non-structural protein NS1, and that pericyte dysregulation may explain the delay between peak serum NS1 levels and the later development of life-threatening vascular dysfunction. NS1’s effects were mediated through contact-mediated and contact-independent signalling between pericytes and endothelial cells. Importantly, NS1 disrupted microvascular spheroids that recapitulate in vivo multicellular interactions, and induced liver-specific microvascular pathology that may explain liver dysfunction in patients. Finally, we showed for the first time that DENV NS1 impairs lymphatic vessel function. By developing a more complete picture of the human microvasculature, we therefore show that dengue fluid accumulation is driven by both leakage from capillaries and impaired reabsorption by lymphatics. Furthermore, pericytes and lymphatics play a crucial role in amplifying dengue microvascular leakage and their dysfunction contributes to tissue-specific pathology, which may open new therapeutic and prognostic opportunities.