Résumé : Background Spontaneous bacterial peritonitis (SBP) is a common complication among cirrhotic patients. Guidelines recommend third-generation cephalosporins (3GCs) as empiric antibiotic therapy (EAT) of SBP. Recently, a broad-spectrum EAT was shown to be more effective than cephalosporins in the treatment of nosocomial spontaneous bacterial peritonitis (N-SBP); however, the prevalence of 3GCs-resistant bacteria is high in the nosocomial setting and broad-spectrum EAT cannot be used in all cases of SBP. Aim The aim of this study was to evaluate the 3GCs resistance distribution between N-SBP and community-Acquired spontaneous bacterial peritonitis (CA-SBP) to clarify whether 3GCs are still an effective therapeutic intervention for CA-SBP. Methods We searched for studies that reported the aetiology of SBP and the resistance profile of both gram-positive and gram-negative bacteria in MEDLINE and Google Scholar databases (since 1 January 2000 to 30 April 2017). A meta-Analysis was carried out to estimate the risk difference [relative risk (RR) and 95% confidence intervals (CIs)] for 3GCs resistance in N-SBP and CA-SBP. Heterogeneity was assessed using the I 2-test. Results A total of eight studies were included, including 1074 positive cultures of ascitic fluid in cirrhotic patients; 462 positive cultures were from N-SBP and, among these, 251 (54.3%) were 3GCs resistant. Six hundred and twelve positive cultures were from CA-SBP and, among these, 207 (33.8%) were 3GCs-resistant SBP. A pooled RR of 3GCs resistance in N-SBP compared with CA-SBP showed a significant difference (RR=1.67, 95% CI: 1.14-2.44; P=0.008). We carried out two subgroup analyses: the first according to the median year of study observation (before vs. since 2008) and the second according to the country of the study (China vs. others). The studies carried out before 2008 (327 SBP-positive culture) showed a significantly higher risk for 3GCs-resistant strains in N-SBP compared with CA-SBP (RR=2.36, 95% CI: 1.39-3.99; P=0.001), whereas this was not found in SBP acquired after 2008 (RR=1.24, 95% CI: 0.83-1.84; P=0.29). N-SBP occurring in China had no significantly higher risk for 3GCs-resistant strains compared with CA-SBP (RR=1.44, 95% CI: 0.87-2.37; P=0.16). Conclusion Our findings suggest that although the pooled RR of 3GCs resistance in N-SBP compared with CA-SBP show that 3GCs are still an effective option for the treatment of CA-SBP, the subanalysis of studies that enroled patients in the last decade did not show a significant higher RR of 3GCs resistance in N-SBP compared with CA-SBP. Therefore, in centres where local patterns of antimicrobial susceptibility (with low rates of 3GCs resistance) are not available, 3GCs should not be used initially for CA-SBP treatment. Future studies are needed to confirm this trend of 3GCs resistance.