Objective To investigate the accuracy and feasibility of brachial artery peak velocity variation (ΔVpeakbrach) and inferior vena cava variability (VIVC) as indicators of fluid responsiveness in critically ill patients. Methods A single-center prospective observation was conducted. The patients on mechanical ventilation with spontaneously breathing admitted to Department of Critical Care Medicine of the Second Affiliated Hospital of Kunming Medical University from June 2013 to August 2015 were enrolled. The patients were diagnosed as severe sepsis or sepsis shock. The peak velocity in brachial artery and diameter of the inferior vena cava at the end of inspiration and expiration was measured by bedside portable ultrasonic machine, and then ΔVpeakbrach and VIVC were calculated. The hemodynamic parameters were collected at baseline and after volume expansion (VE). The stroke volume (SV) was measured by pulse-indicated continuous cardiac output (PiCCO). Patients were classified as responders or non-responders according to the variation of SV (ΔSV) increased ≥ 15% or not after VE. Receiver operating characteristic curve (ROC) was plotted to evaluate the sensitivity and specificity of ΔVpeakbrach and VIVC in predicting volume responsiveness. Results Among 58 patients after VE, 32 patients were defined as responders and the rest 26 were defined as non-responders.There were no differences in gender, age, acute physiology and chronic health evaluation Ⅱ (APACHE Ⅱ) score, dose of vasoactive agent, ventilator parameters and infection site. Compared with baseline hemodynamic parameters, heart rate (HR) was decreased (bpm: 95±18 vs. 103±21), and systolic blood pressure (SBP) was increased [mmHg (1 mmHg = 0.133 kPa): 92±8 vs. 80±7] after VE in responders; central venous pressure (CVP) was increased after VE in non-responders (mmHg: 11±4 vs. 8±3, all P < 0.05). The ΔVpeakbrach [(15.4±4.3)% vs. (11.2±3.5)%] and VIVC [(18.6±4.1)% vs. (14.3±3.6)%] in responders were significantly increased as compared with those of non-responders (both P < 0.05). The area under ROC curve (AUC) of ΔVpeakbrach for predicting volume responsiveness was 0.816. When the cut-off value of ΔVpeakbrach was ≥ 13.3%, the sensitivity was 71.9%, and the specificity was 80.8%. AUC of VIVC for predicting volume responsiveness was 0.733. When the cut-off value of VIVC was ≥ 19.25%, the sensitivity was 53.1%, and the specificity was 88.5%. Conclusion ΔVpeakbrach and VIVC are reliable indicators for predicting volume responsiveness in critical patients.