OBJECTIVE: To explore the relationship between cholesterol levels and the adrenal cortisol response to synacthen in critically ill patients. DESIGN: Prospective observational study. PATIENTS: Critically ill patients with multiple organ dysfunction syndrome (MODS) with possible adrenal dysfunction defined as unexplained hypotension, ongoing inotropic support, unexplained fever, unexplained hyponatraemia or a combination of these symptoms. MEASUREMENTS: HDL-cholesterol levels (HDL), total cholesterol levels (TC), and triglycerides (TG) before administration of synacthen. LDL-cholesterol was calculated using the Friedewald formula. Basal cortisol and response to 250 microg synacthen intravenously was measured. A cortisol rise of 0.25 micromol/l in a 30-min or 60-min blood sample after synacthen infusion was defined as a proper adrenal response. RESULTS: Patients with a proper response to synacthen showed higher HDL-cholesterol levels than patients without that response ( P=0.02). Severity of disease as measured by APACHE II or SOFA was not a confounder. LDL-cholesterol levels were extremely low in both responders and non-responders and were not associated with the absolute rise in cortisol. In linear and logistic regression analysis HDL-cholesterol was the sole predictor of cortisol response. CONCLUSIONS: Adrenal cortisol response to a "classic" 250-microg synacthen test relates in critically ill patients to HDL-cholesterol levels. LDL and TC levels did not show such a relation. These findings are in concordance with known biochemical pathways of cortisol production.