OBJECTIVE To evaluate the value of procalcitonin (PCT) detection in the diagnosis of local infection and sepsis. METHODS PCT, C-reactive protein (CRP), white blood cell count (WBC), neutrophil ratio (neu%) and lymphocyte ratio (lym%) were measured in patients with negative or positive blood culture test. The receiver operating characteristic (ROC) curves were constructed for PCT CRP, WBC, neu%, lym%, and the diagnostic model using SPSS software. Based on the binary logistic regression model, the predictors or probabilities were obtained and applied to establish the empirical and binormal model of the ROC curves to compare the area under the curve (AUC). RESULTS A highly significant difference in PCT concentrations was noted between the two groups (chi(2)=52.52, P<0.001), and the diagnostic criteria at <2 of the ROC curves resulted in the greatest Youden index with a sensitivity of 63.3% and specificity of 86.8%. The AUC of PCT, CRP, WBC, neu% and lym% were 0.700, 0.765, 0.636, 0.618 and 0.648, respectively; the combined predicted ROC AUC was 0.776. The maximum Youden index was acquired at the optimal cutoff point of 0.566 with a diagnosis sensitivity and specificity of 63.8% and 84.7%, respectively. CONCLUSIONS The PCT level is a valuable predictor for a rapid and reliable early diagnosis of sepsis. The diagnostic model based on the laboratory parameters, using the combined predictors of PCT, CRP and lym%, can be a useful means for predicting early-onset sepsis.