Assessment of the cardiac output (CO) is usually performed with invasive techniques requiring specialized equipment in the intensive care unit (ICU). With TEL-O-GRAPH (TG), CO can be derived from the oscillometrically obtained brachial pulse wave during the measurement of brachial blood pressure. CO and stroke volume (SV) determinations with TG were compared with transpulmonary thermodilution measurements with the PICCO system (PICCO) in 38 haemodynamically unstable ICU patients with a total of 84 comparison measurements performed. SV (33.3 ± 9.0 ml/m2 vs. 44.3 ± 14.4 ml/m2, p < 0.001) and CO (2.7 ± 0.5 l/min/m2 vs. 3.8 ± 1.2 l/min/m2, p < 0.001) were underestimated significantly with TG and oscillometric brachial systolic blood pressure (BP) was significantly lower and diastolic BP significantly higher than invasive femoral artery pressure. A linear correlation was found between CO dimension and CO underestimation with TG. Correct tracking of CO changes with a fluid challenge was possible in 69.5% of measurements. Oscillometric noninvasive CO is possible in the ICU, but accuracy and precision of this new method are lacking. Implementation of a correction factor accounting for the linear increase in CO underestimation observed with increasing CO could improve CO assessment with TG in haemodynamically unstable patients.