Analysis of breath acetone could be useful in the Intensive Care Unit (ICU) setting to monitor evidence of starvation and metabolic stress. The aims of this study were to examine the relationship between acetone concentrations in breath and blood in critical illness, to explore any changes in breath acetone concentration over time and correlate these with clinical features. Consecutive patients, ventilated on controlled modes in a mixed ICU, with stress hyperglycaemia requiring insulin therapy and/or new pulmonary infiltrates on chest radiograph were recruited. Once daily, triplicate end-tidal breath samples were collected and analysed off-line by selected ion flow tube mass spectrometry (SIFT-MS). Thirty-two patients were recruited (20 males), median age 61.5 years (range 26-85 years). The median breath acetone concentration of all samples was 853 ppb (range 162-11 375 ppb) collected over a median of 3 days (range 1-8). There was a trend towards a reduction in breath acetone concentration over time. Relationships were seen between breath acetone and arterial acetone (rs = 0.64, p < 0.0001) and arterial beta-hydroxybutyrate (rs = 0.52, p < 0.0001) concentrations. Changes in breath acetone concentration over time corresponded to changes in arterial acetone concentration. Some patients remained ketotic despite insulin therapy and normal arterial glucose concentrations. This is the first study to look at breath acetone concentration in ICU patients for up to 8 days. Breath acetone concentration may be used as a surrogate for arterial acetone concentration, which may in future have a role in the modulation of insulin and feeding in critical illness.