B-type natriuretic peptide is known to predict outcome in congestive cardiac failure and myocardial infarction. We aimed to determine whether measurement of B-type natriuretic peptide would predict hospital mortality in patients admitted to an intensive care unit. We conducted a prospective observational cohort study in 78 consecutive patients. Demographics, clinical details and clinical outcomes were recorded. Admission and 24 h B-type natriuretic peptide and cardiac troponin I levels were measured. B-type natriuretic peptide and cardiac troponin I levels taken on intensive care admission and 24 h after admission did not accurately predict hospital mortality for all patients, including patients with severe sepsis or septic shock (all p > 0.05). B-type natriuretic peptide levels were higher in patients with severe sepsis and septic shock (p = 0.02), in patients > or = 65 years (p = 0.04) and in patients with raised creatinine > or = 110 micromol.l(-1) (p = 0.02). We concluded that B-type natriuretic peptide, measured soon after admission to intensive care, does not usefully predict outcome after intensive care.