Objective To investigate the value of peripheral blood for the prognosis of patients withbloodstream infection. Methods A retrospective analysis of patients with bloodstream infection was conducted inthe intensive care unit (ICU) of Mianyang Central Hospital of Sichuan from January 2012 to October 2016. Accordingto the 28-day survival, the patients were divided into survival group and death group. The white blood cell (WBC),neutrophils count (NEU), lymphocyte count (LYM), neutrophil/lymphocyte ratio (NLR), monocyte count (MO), eosinophilcount (EO), basophil count (BA), hemoglobin (Hb), platelet count (PLT) and procalcitonin (PCT) in peripheral bloodwere recorded when patients were diagnosed with blood infection. Receiver operating characteristic curve (ROC),Kaplan-Meier survival analysis and Cox regression were used to evaluate the value of these risk factors for predictingthe outcome. Results 305 patients were enrolled. 182 patients survived while 123 patients died during the 28-dayperiod. ① There was no significant difference in gender, age and comorbidities between the two groups. There was nosignificant difference in infection rate between the two groups except for fungal infection rate. The fungal infection ratein the death group was significantly higher than that in the survival group (9.8% vs. 3.3%, P = 0.019). ② The LYM,MO, EO and PLT in the death group were significantly lower than those in the survival group [LYM (×109/L):0.58 (0.29, 0.93) vs. 0.76 (0.44, 1.23), MO (×109/L): 0.47 (0.19, 0.80) vs. 0.58 (0.30, 0.94), EO (×109/L):0.00 (0.00, 0.01) vs. 0.03 (0.01, 0.09), PLT (×1012/L): 89 (47, 148) vs. 126 (82, 186), all P < 0.05]. The NLR in the death group was significantly higher than that in the survival group [17.09 (7.60, 33.51) vs. 12.86 (6.51, 24.85), P < 0.05]. There was no significant difference in the WBC, NEU, BA, Hb and PCT between the two groups. ③ It was shown by ROC curve analysis that the maximum area under the ROC curve (AUC) of EO was 0.755. When the best cut-off value was 0.015×109/L as a predictor of death in 28 days, the sensitivity was 80.3%, and specificity was 64.7%. ④ It was shown by survival analysis that the 28-day survival rate in the patients with EO < 0.015×109/L was significantly lower than that of patients with EO > 0.015×109/L [38.3% (62/162) vs. 83.9% (120/143), χ 2 = 56.999, P = 0.000]. ⑤ It was shown by Cox regression that EO was the independent factor for 28-day survival (β = 1.466, χ 2 = 39.535, P = 0.000). Risk of death was 4.331 times greater in patients with EO < 0.015×109/L than in those with EO > 0.015×109/L [hazard ratio (HR) = 4.331, 95% confidence interval (95%CI) = 2.743-6.840]. Conclusions Compared to other parameters in peripheral blood, EO has the best correlation with the prognosis of bloodstream infection. EO is the independent prognostic predictor for 28-day survival.