Background <p>Myocardial injury is common in sepsis and associated with increased mortality, but its underlying mechanisms remain incompletely understood. Coronary microvascular dysfunction (CMD) has been proposed as a contributor, although in vivo evidence is limited. We characterised coronary microvascular function in patients with sepsis and myocardial injury and its relationship with cardiac troponin release.</p> Methods <p>Consecutive adults with sepsis (Sepsis-3 criteria) and myocardial injury (hs-cTnT ≥ 15 ng/L) were prospectively enrolled between June 2019 and December 2024. Patients underwent coronary angiography, invasive thermodilution-based assessment of coronary microvascular function, and transthoracic echocardiography after clinical stabilisation. CMD was defined as an index of microcirculatory resistance (IMR) &gt; 25 and/or microvascular resistance reserve (MRR) ≤ 3. Associations between hs-cTnT concentrations and coronary microvascular indices, obstructive coronary artery disease (CAD), and echocardiographic variables were assessed using regression analyses. Coronary microvascular indices were compared with those of age-, sex-, and CAD-matched patients with chronic coronary syndrome (CCS) using mixed-effects models.</p> Results <p>Fifty-five patients underwent coronary angiography and 49 completed invasive coronary microvascular assessment. Obstructive CAD was identified in 12/55 (22%). CMD was present in 30/49 (61%). Among these, 8/49 (16%) had elevated IMR only, 9/49 (18%) had functional CMD (MRR ≤ 3 and IMR ≤ 25), and 13/49 (27%) had structural CMD (MRR ≤ 3 and IMR &gt; 25). Restricted cubic spline analyses demonstrated no evidence of associations between hs-cTnT and IMR (overall <i>P</i> = 0.899; non-linearity <i>P</i> = 0.687) or MRR (overall <i>P</i> = 0.987; non-linearity <i>P</i> = 0.954). CMD was associated with a higher prevalence of ventriculo-arterial uncoupling, right ventricular systolic dysfunction, and impaired right ventricular-pulmonary arterial coupling. Patients with sepsis demonstrated reduced coronary microvascular vasodilatory capacity compared with matched CCS controls (MRR 3.2 [IQR 2.4–4.5] vs. 4.0 [2.7–6.2]). IMR was similar between groups.</p> Conclusions <p>CMD was common and heterogeneous and exhibited distinct phenotypes in patients with sepsis and myocardial injury. Measures of coronary microvascular function were not associated with troponin release. Previously unrecognised obstructive CAD was identified in 22% of patients, supporting consideration of underlying CAD in patients with sepsis and myocardial injury.</p> Trial registration <p>ClinicalTrials.gov identifier NCT06294730.</p> Graphical abstract <p></p>