Abstract: Objective To study the therapeutic effect of phages on extensively drug-resistant Acinetobacter baumannii-induced sepsis in mice. Methods (1) Sixty BALB/c mice were divided into blank control group, sepsis control group, antibiotics treatment group, phage treatment group, and phage control group according to the random number table, with 12 mice in each group. Mice in blank control group were intraperitoneally (the same injection position below) injected with 1 mL normal saline. Mice in sepsis control group, antibiotics treatment group, and phage treatment group were injected with 1 mL extensively drug-resistant Acinetobacter baumannii (the strain was isolated from the blood of a severely burned patient hospitalized in our unit) in the concentration of 5×10⁷ colony-forming unit/mL to reproduce sepsis model. Two hours later, mice in sepsis control group, antibiotics treatment group, and phage treatment group were injected with 1 mL saline, 1 mg/mL imipenem/cilastatin, and 1×10⁸ plaque-forming unit (PFU)/mL phages screened based on above-mentioned Acinetobacter baumannii (the same phages below) respectively. Mice in phage control group were injected with 1 mL phages in the titer of 1×10⁸ PFU/mL. The injection was performed continuously for 7 days in each living mouse, and the survival situation of mice was observed each day to calculate the survival ratio in one week. (2) Another 60 BALB/c mice were grouped and treated as in experiment (1), and the injection was performed continuously for 5 days in each living mouse. On experiment day 2, 4, and 6, 3 mice from each group were selected (if the number of survived mouse in any group was less than 3 at sample collecting, all the survived mice were selected), and blood was drawn to determine white blood cell count (WBC, with 3 samples at each time point in each group). On experiment day 2, blood was drawn from the mice that had their blood taken earlier for bacterial culture, and lung, liver, kidney, and spleen tissue was collected from the same mice. The tissue samples were added to the LB solid medium after being homogenized and diluted for bacterial culture. The content of bacteria was calculated after the bacterial colony number was counted. Data were processed Wilcoxon rank sum test, one-way analysis of variance, LSD test and Kruskal-Wallis rank sum test. Results (1) On experiment day 7, there were 12, 8, 10, and 12 mice survived in blank control group, antibiotics treatment group, phage treatment group, and phage control group respectively, while no mouse survived in sepsis control group. Compared with that in sepsis control group, the survival ratio of mice was significantly higher in the other four groups (with Z values from 55.635 to 106.593, P values below 0.05). The survival ratio of mice in phage treatment group was slightly higher than that in antibiotics treatment group, without statistically significant difference (Z=2.797, P>0.05). (2) On experiment day 2, WBC data of mice in blank control group, phage treatment group, and phage control group were close[respectively (5.60±0.94)×10⁹/L, (5.16±0.36)×10⁹/L, and (5.26±1.89)×10⁹/L], all significantly lower than the datum in sepsis control group[(8.64±0.64)×10⁹/L, P<0.05 or P<0.01], and the WBC data in the latter two groups were significantly lower than the datum in antibiotics treatment group[(7.80±1.76)×10⁹/L, with P values below 0.05]. On experiment day 4, WBC data of mice in antibiotics treatment group, phage treatment group, and phage control group were close, all significantly lower than the datum in blank control group (P<0.05 or P<0.01), and WBC data in the above-mentioned four groups were all lower than the datum in sepsis control group (with P values below 0.01). On experiment day 6, there was no statistically significant difference in WBC among blank control group, antibiotics treatment group, phage treatment group, and phage control group (χ²=4.128, P>0.05). On experiment day 2, respectively 12, 7, and 2 mice were detected as blood bacterial culture-positive in sepsis control group, antibiotics treatment group, and phage treatment group, while no positive result was detected in the other two groups. Positive ratios of blood bacterial culture of mice in blank control group, phage treatment group, phage control group were significantly lower than the ratio in sepsis control group (with χ² values from -30.000 to 30.000, P values below 0.01). Positive ratio of blood bacterial culture of mice in antibiotics treatment group was significantly higher than that in blank control group or phage control group (with χ² values respectively 17.500 and -17.500, P values below 0.05). On experiment day 2, except for the kidney tissue of mice in phage treatment group, the bacteria load in each viscus of mice in blank control group, phage treatment group, and phage control group was significantly lower than that in sepsis control group (with χ² values from -9.000 to 9.000, P values below 0.01). The bacteria load in kidney of mice in antibiotics treatment group was significantly higher than that in blank control group or phage control group (with χ² values respectively -7.500 and 7.500, P values below 0.05). Conclusions Phages can significantly improve survival ratio, control inflammation response, and effectively clean bacteria in lung, liver, spleen, and kidney in treating extensively drug-resistant Acinetobacter baumannii-induced sepsis in mice.