Recent work suggests that increased plasma concentrations of cardiac troponin I (cTnI) are common in critically ill patients and are associated with poor outcome. We measured the frequency of increased plasma cTnI concentrations during patients' stay in a mixed medical/surgical intensive care unit (ICU) and compared our findings with hospital mortality. Basic details, organ support, and hospital mortality were recorded for all patients treated in ICU during a 6 month period. cTnI concentrations were sampled daily for all patients, using 0.04 µg litre−1 as the upper limit of normal, and 0.12 µg litre−1 as an additional stratification point. Of 663 patients, 54% were male, with a mean (sd) age of 60 (18) yr, 65% were surgical patients, and the median Acute Physiology and Chronic Ill Health II (APACHE II) score was 15 (inter-quartile range 12–20). Increased cTnI concentrations were found in 345 patients (52%) while in ICU. One hundred and twenty patients (18%) died in hospital. cTnI concentration >0.04 µg litre−1 was associated with reduced odds of hospital survival, independent of age, medical admission, unplanned admission, APACHE II score, mechanical ventilation, and haemofiltration (adjusted odds ratio 0.25, 95% confidence interval 0.08–0.75, P=0.014). Stratification by the degree of cTnI increase revealed an incremental trend towards a lower odds of hospital survival, including for patients with ‘minor' elevations of cTnI (0.05–0.12 µg litre−1). Increased serum cTnI concentrations during ICU stay independently predicts hospital mortality, even when the threshold is low. We found a trend towards an association between ‘minor' elevations in cTnI and higher in-hospital mortality.