This study aimed to assess the viability of using the venous-to-arterial carbon dioxide partial pressure difference (P(v-a)CO2) to predict clinical worsening of septic shock, depending on central venous oxygen saturation (ScvO2). The prospective, observational, multicentric study conducted in three intensive care units (ICUs) included all patients with a septic shock episode during the first 6 h, with 122 patients assessed. Clinical worsening was defined as an increase of sequential organ failure assessment (SOFA) scores ≥1 (ΔSOFA ≥1) within 2 days. To assess the ability of P(v-a)CO2 to predict clinical worsening, univariate and multivariate analyses were performed according to ΔSOFA. A receiver-operating characteristic (ROC) analysis was used to confirm model predictions. Associations between P(v-a)CO2 and mortality were explored using correlations. Using multivariate analyses, two independent factors associated with ΔSOFA at least 1 were identified: an averaged 6-h value of lactate concentration (Lac [1-6]) (odds ratios [ORs], 2.43 [95% confidence interval, CI, 1.20-4.89]; P = 0.013) and an averaged 6-h value of P(v-a)CO2 (P(v-a)CO2 [1-6]) (OR, 1.49 [95% CI, 1.04-2.15]; P = 0.029). ROC analysis confirmed that Lac [1-6] and P(v-a)CO2 [1-6] were significantly associated with ΔSOFA at least 1, whereas ScvO2 [1-6] was not. Finally, ΔSOFA at least 1 was associated with higher 28-day (76% vs. 10%, P = 0.001) and ICU (83% vs. 12%, P = 0.001) mortality rates, which were higher in patients with P(v-a)CO2 [1-6] more than 5.8 mmHg (57% vs. 33%; P = 0.012). In conclusion, P(v-a)CO2 may help predict outcomes for septic shock patients regardless of ScvO2 values.