The definition of sepsis was updated to sepsis-3 in February 2016. However, the performance of the previous and new definition of sepsis remains unclear in China. This was a retrospective multicenter study in six intensive care unit (ICUs) from five university-affiliated hospitals to compare the performance between sepsis-1 and sepsis-3 in China. From May 1, 2016 to June 1, 2016, 496 patients were enrolled consecutively. Data were extracted from the electronic clinical records. We evaluated the performance of sepsis-1 and sepsis-3 by measuring the area under the receiver operating characteristic curves (AUROC) to predict 28-day mortality rates. Of 496 enrolled patients, 186 (37.5%) were diagnosed with sepsis according to sepsis-1, while 175 (35.3%) fulfilled the criteria of sepsis-3. The AUROC of systemic inflammatory response syndrome (SIRS) is significantly smaller than that of sequential organ failure assessment (SOFA) (0.55 [95% confidence interval, 0.46–0.64] vs. 0.69 (95% confidence interval, 0.61–0.77], P = 0.008) to predict 28-day mortality rates of infected patients. Moreover, 5.9% infected patients (11 patients) were diagnosed as sepsis according to sepsis-1 but not to sepsis-3. The APACHE II, SOFA scores, and mortality rate of the 11 patients were significantly lower than of patients whose sepsis was defined by both the previous and new criteria (8.6±3.5 vs. 16.3±6.2, P = < 0.001; 1 (0–1) vs. 6 (4–8), P = <0.001; 0.0 vs. 33.1%, P = 0.019). In addition, the APACHE II, length of stay in ICU, and 28-day mortality rate of septic patients rose gradually corresponding with the raise in SOFA score (but not the SIRS score). Sepsis-3 performed better than sepsis-1 in the study samples in ICUs in China.