We planned a study to assess the safety, feasibility, and efficacy of the Impella micro-axial blood pump in patients with cardiogenic shock. From January 2001 to September 2002 inclusive, 16 patients in cardiogenic shock (maximal inotropic support and with IABP in 11 cases) underwent left ventricle unloading with the Impella pump. 6 were placed via the femoral artery (patients in the coronary care unit) and 10 directly through the aorta (postcardiotomy heart failure). In three patients, the device was used in combination with ECMO. Mean age was 60 years (range 43 - 75), 11 were male. A stable pump flow of 4.24 +/- 0.28 l/min was reached (3.3 +/- 1.9 l/min in patients with ECMO and Impella). Mean blood pressure before Impella) support was 57.4 +/- 13 mmHg, which increased to 74.9 +/- 13 mmHg after 6 hours and 80.6 +/- 17 mmHg (p = 0.003) after 24 hours. Cardiac output increased from 4.1 +/- 1.3 l/min to 5.5 +/- 1.3 (p = 0.003) and 5.9 +/-1.9 l/min (p = 0.01) at 6 and 24 hours. Mean pulmonary wedge pressure decreased from 29 +/- 10 mmHg to 17 +/- 5 mmHg and 18 +/- 7 mmHg at 6 (p = 0.04) and 24 hours. Blood lactate levels decreased significantly after 6 hours of support (from 2.7 +/- 1 to 1.3 +/- 0.5 mmol/l, p = 0.004). Device-related complications included three sensor failures (no clinical action), one pump displacement (replacement) and six incidences of haemolysis (peak free plasma haemoglobin > 100 mg/dl, no clinical action). Eleven patients (68 %) were weaned, six (37 %) survived. Left ventricular unloading with the Impella pump via the transthoracic or femoral approach is feasible and safe. Support led to a decrease in pulmonary capillary wedge pressure, increase in cardiac output and mean blood pressure, and improved organ perfusion in patients with severe cardiogenic shock.