S100β and neuron-specific enolase (NSE) are brain injury biomarkers, mainly used in brain trauma, cerebral stroke and hypoxic ischemia encephalopathy. The aim of this study was to study the clinical significance of serum S100β and NSE in diagnosing sepsis-associated encephalopathy (SAE) and predicting its prognosis. This was a prospective and observational study. Clinical data of septic patients were collected within 24 h after ICU admission from May 2012 to April 2013. We evaluated the level of consciousness twice per day. SAE was defined as cerebral dysfunction in the presence of sepsis that fulfilled the exclusion criteria. The infection biochemical indicators, Glasgow coma scale (GCS) score, acute physiology and chronic health evaluation score II, serum NSE and S100β were newly measured or evaluated for SAE patients. Finally, hospital mortality, bacteriological categories, length of ICU stay and length of hospital stay were also recorded for all enrolled patients. The data was analyzed with the Chi square test, two-sample t test or Mann-Whitney U test between two groups. The correlation between two factors was analyzed using the Pearson or Spearman analysis. Receiver operating characteristic (ROC) curves were used to determine the ability of S100β and NSE in diagnosing SAE and predicting the hospital mortality. In addition, cut-off points were obtained from the curves to determine the highest sum of sensitivity and specificity. Of 112 enrolled patients, 48 patients were diagnosed with SAE. The serum S100β and NSE concentrations in SAE patients were both significantly higher than in non-SAE patients 0.306 (IQR 0.157-0.880) μg/L vs. 0.095 (IQR 0.066-0.177) μg/L, 24.87 (IQR 31.73-12.73) ng/mL vs. 15.49 (IQR 9.88-21.46) ng/mL, P < 0.01]. GCS scores were related more closely to S100β than NSE (-0.595 vs. -0.337). S100β levels of 0.131 μg/L diagnosed SAE with 67.2% specificity and 85.4% sensitivity in the ROC curve, the area under the curve was 0.824 (95% confidence interval 0.750-0.898). NSE levels of 24.15 ng/mL diagnosed SAE with 82.8% specificity and 54.2% sensitivity, and the area under the curve was 0.664 (95 % confidence interval 0.561-0.767). In addition, the area under the curve for S100β for predicting hospital mortality was larger than for NSE (0.730 vs. 0.590). Serum S100β concentrations in SAE patients were significantly higher than in non-SAE patients. These may be related to the severity of SAE and may predict the outcome of sepsis. The efficacy and sensitivity of serum S100β in diagnosing SAE were high, but it had a low specificity. Moreover, compared to NSE, serum S100β was better for both diagnosing SAE and predicting the outcome of sepsis.