Objective To explore the expression differences of procalcitonin (PCT) in different infection sites and bacterial strains, and to provide the evidence for early differential diagnosis of infectious diseases with PCT as a biomarker. Methods The patients with various kinds of infections diagnosed in West China Hospital of Sichuan University between January 2012 and June 2016 were retrospectively included. The expression differences of PCT in various infection sites and bacterial strains were analyzed. Results A total of 1 005 patients were include in this study, including 259 with systemic infection and 746 with local infection. The median PCT level in the systemic infection group was higher than that in the local infection group (8.57 vs. 0.10 ng/mL, P<0.05). In the 779 patients with pulmonary infection, the median PCT level of the patients with sepsis caused by pulmonary infection was higher than that of the ones without sepsis (4.61vs. 0.10 ng/mL, P<0.05), and the median PCT level of the patients with positive sputum culture was higher than that of the ones with negative sputum culture (0.28vs. 0.08 ng/mL, P<0.05). In the 48 patients with urinary tract infection, the median PCT level of the patients with sepsis caused by urinary tract infection was higher than that of the ones without sepsis (12.00vs. 0.42 ng/mL, P<0.05), and the median PCT level of the patients with complicated urinary tract infection was higher than that of the patients with simplex urinary tract infection (19.15vs. 5.02 ng/mL, P<0.05). In the 259 patients with systemic infection, the median PCT level of the patients with infective shock was higher than that of the ones without infective shock (40.26vs. 3.83 ng/mL, P<0.05); the mean PCT level of patients with infection of Gram-negative bacteria, Gram-positive bacteria and fungi was 13.66, 0.99, and 3.30 ng/mL with a significant difference (P<0.05). Conclusion The PCT level has unique advantages in identifying different sites of the infection, early diagnosing complicated urinary tract infection, and evaluating the severity of infection, which could provide evidence in early identification for sepsis caused by various kinds of infectious pathogens.