The present pilot study investigated whether respiratory variation in subclavian vein (SCV) diameters correlates with fluid responsiveness in mechanically ventilated patients. Monocentric, prospective clinical study on fluid responsiveness in adult sedated, mechanically ventilated ICU patient, monitored with the PiCCO™ system (Pulsion Medical System, Germany), and requiring a fluid challenge (FC). A 10-min fluid bolus of 500 mL of 0.9% saline was administered. Cardiac output (CO) and dynamic parameters [stroke volume variation (SVV) and pulse pressure variation (PPV)] measured by transpulmonary thermodilution and pulse contour analysis (PiCCO™) as well as classical hemodynamic parameters were recorded at baseline and after FC. Fluid responsiveness was described as an increase in CO of ≥ 15%. Ultrasound measurements obtained in the subclavian long-axis view were used to calculate the SCVvariability index. A cut-off value for SCV variation for the prediction of fluid responsiveness was determined using receiver operating curve (ROC) analysis. Nine of 20 FCs (45%) induced an increase in CO of ≥ 15%. At baseline, the SCVvariability index was greater in responders than in non-responders (34.0 ± 21.4 vs. 9.0 ± 5.5; p = 0.0005). Diagnostic performance for the SCVvariability index revealed a cut-off value of 14 with a sensitivity of 100% [Confidence interval (CI) 95% (90; 100)] and a specificity of 82% [CI 95% (48; 98)] for the prediction of fluid responsiveness. Other parameters, such as SVV and PPV, could not predict fluid responsiveness. The correlation coefficient between CO variation and the SCVvariability index was 0.73 (p < 0.001). The SCVvariability index was a reliable, non-invasive parameter for the prediction of fluid responsiveness at the bedside of mechanically ventilated, critically ill patients in this pilot study.