Objective To investigate the value of central venous-to-arterial carbon dioxide difference/arterial-to-venous oxygen difference ratio [P(cv-a)CO2/C(a-cv)O2] in predicting oxygen metabolism after fluid resuscitation in patients with septic shock. Methods A prospective observational study was carried out on septic shock patients admitted in the intensive care unit of Nanjng Drum Tower Hospital from November 2013 to April 2014. All patients underwent fluid challenge (300 ml saline for 20 min, rapid intravenous infusion). The patients were divided into a fluid responded group (ΔCI≥10%) and a fluid unresponded group (ΔCI＜10%), according to the change of cardiac output index (ΔCI) after fluid challenge. Then the patients were divided into two subgroups in the fluid responded group, namely a ΔVO2≥10% group and a ΔVO2＜10% group, according to the change of VO2 (ΔVO2). Cardiac output index (CI) were determined by pulse indicator continuous cardiac output (PICCO). Hemoglobin, arterial carbon dioxide (PaCO2), arterial oxygen (PaO2), arterial oxygen saturation (SaO2), arterial blood lactate, central venous carbon dioxide (PcvCO2), central venous oxygen (PcvO2) and central venous oxygen saturation (ScvO2) were measured by blood gas analysis. P(cv-a)CO2/C(a-cv)O2 and oxygen consumption (VO2) were calculated. P(cv-a)CO2/C(a-cv)O2 before and after fluid challenge was compared between two subgroups. Results Fluid challenges were performed in 23 instances in 18 patients, among which 17 instances were defined as the fluid responded group. Compared with the fluid unresponded group, P(cv-a)CO2/C(a-cv)O2, arterial lactate and ScvO2 had no significant difference [P(cv-a)CO2/C(a-cv)O2](mm Hg/ml): 2.05±0.75vs. 1.58±0.67; arterial lactate (mmol/l): 3.78±2.50vs. 3.26±2.42; ScvO2(%): 73.71±9.64vs. 70.30±12.01,P＞0.05] in the fluid responded group before resuscitation. In the fluid responded group, there were 10 instances in the ΔVO2≥10% group and 7 instances in the ΔVO2＜10% group. P(cv-a)CO2/C(a-cv)O2 (mm Hg/ml) was significantly higher in the ΔVO2≥10% group before resuscitation compared with the ΔVO2＜10% group (2.43±0.73vs. 1.51±0.37,P＜0.01). Lactate (mmol/l) was also higher in the ΔVO2≥10% group before resuscitation (4.53±2.52vs. 1.46±0.82,P＜0.01). ScvO2 (%) had no significant difference between two groups (70.79±9.15vs. 72.13±13.42,P＞0.05). The areas under ROC curve (AUCs) of P(cv-a)CO2/C(a-cv)O2, lactate and ScvO2 for predicting ΔVO2≥10% were 0.843, 0.921, and 0.529, respectively. The sensitivity and specificity of P(cv-a)CO2/C(a-cv)O2≥1.885 mm Hg/ml for predicting ΔVO2≥10% after fluid resuscitation were 70% and 86%, respectively. Conclusion For septic shock patients with fluid responsiveness, P(cv-a)CO2/C(a-cv)O2 can predict oxygen metabolism after fluid resuscitation and can be used as a reliable parameter to guide fluid resuscitation.