Abstract Background COVID-19 is primarily a respiratory disease; however, there is also evidence that it causes endothelial damage in the microvasculature of several organs. The aim of the present study is to characterize in vivo the microvascular reactivity in peripheral skeletal muscle of severe COVID-19 patients. Methods This is a prospective observational study carried out in Spain, Mexico and Brazil. Healthy subjects and severe COVID-19 patients admitted to the intermediate respiratory (IRCU) and intensive care units (ICU) due to hypoxemia were studied. Local tissue/blood oxygen saturation (StO 2 ) and local hemoglobin concentration (THC) were non-invasively measured on the forearm by near-infrared spectroscopy (NIRS). A vascular occlusion test (VOT), a three-minute induced ischemia, was performed in order to obtain dynamic StO 2 parameters: deoxygenation rate (DeO 2 ), reoxygenation rate (ReO 2 ), and hyperemic response (H AUC ). In COVID-19 patients, the severity of ARDS was evaluated by the ratio between peripheral arterial oxygen saturation (SpO 2 ) and the fraction of inspired oxygen (FiO 2 ) (SF ratio). Results Healthy controls (32) and COVID-19 patients (73) were studied. Baseline StO 2 and THC did not differ between the two groups. Dynamic VOT-derived parameters were significantly impaired in COVID-19 patients showing lower metabolic rate (DeO 2 ) and diminished endothelial reactivity. At enrollment, most COVID-19 patients were receiving invasive mechanical ventilation (MV) (53%) or high-flow nasal cannula support (32%). Patients on MV were also receiving sedative agents (100%) and vasopressors (29%). Baseline StO 2 and DeO 2 negatively correlated with SF ratio, while ReO 2 showed a positive correlation with SF ratio. There were significant differences in baseline StO 2 and ReO 2 among the different ARDS groups according to SF ratio, but not among different respiratory support therapies. Conclusion Patients with severe COVID-19 show systemic microcirculatory alterations suggestive of endothelial dysfunction, and these alterations are associated with the severity of ARDS. Further evaluation is needed to determine whether these observations have prognostic implications. These results represent interim findings of the ongoing HEMOCOVID-19 trial. Trial registration ClinicalTrials.gov NCT04689477 . Retrospectively registered 30 December 2020.