INTRODUCTION: Sepsis has a high mortality despite advances in management. Microcirculatory and endothelial dysfunction contribute to organ failure, and better tools are needed to assess microcirculatory responses to adjunctive therapies. We hypothesised that peripheral arterial tonometry (PAT), a novel user-independent measure of endothelium-dependent microvascular reactivity, would be impaired in proportion to sepsis severity and related to endothelial activation and plasma arginine concentrations. METHODS: Observational cohort study in a 350-bed teaching hospital in tropical Australia. Bedside microvascular reactivity was measured in 85 adults with sepsis and 45 controls at baseline and 2-4 days later by peripheral arterial tonometry. Microvascular reactivity was related to measures of disease severity, plasma concentrations of L-arginine (the substrate for nitric oxide synthase), and biomarkers of endothelial activation. RESULTS: Baseline reactive hyperaemia index (RH-PAT index), measuring endothelium-dependent microvascular reactivity; (mean [95 CI]) was lowest in severe sepsis (1.57 [1.43-1.70]), intermediate in sepsis without organ failure (1.85 [1.67-2.03]) and highest in controls (2.05 [1.91-2.19]); P <0.00001. Independent predictors of baseline RH-PAT index in sepsis were APACHE II score and mean arterial pressure, but not plasma L-arginine or markers of endothelial activation. Low baseline RH-PAT index was significantly correlated with an increase in SOFA score over the first 2-4 days (r = -0.37, P = 0.02). CONCLUSIONS: Endothelium-dependent microvascular reactivity is impaired in proportion to sepsis severity and suggests decreased endothelial nitric oxide bioavailability in sepsis. Peripheral arterial tonometry may have a role as a user-independent method of monitoring responses to novel adjunctive therapies targeting endothelial dysfunction in sepsis.