OBJECTIVES: To evaluate the clinical safety and effectiveness of carbapenems and other alternative antibiotics in urinary tract infection due to Enterobacteriaceae producing extended-spectrum β-lactamase (ESBL). METHODS: We searched 3 international electronic database including Ovid-Medline, Ovid-EMBASE and Cochrane library and 7 domestic electronic database from its inception to April 2015. Two reviewers independently conducted the data extraction and quality assessment using the Risk of Bias Assessment tool for Non-randomized Studies (RoBANS). RESULTS: A total of 6 articles with 499 patients were finally selected for this study and there were no randomized controlled trials. The diseases included urinary tract infection, complicated lower urinary tract infection and nephropyelitis. Three studies had a comparative study design for carbapenems and non-carbapenems. The risk of selection bias and confounding bias assessed to be high or unclear. Analysis of four studies reporting the microbiological success rate showed that there was no statistically significant difference between the carbapenem group and the non-carbapenem group (OR=1.56, 95% CI 0.44-5.56, I=59%). In the subgroup analysis, both empirical therapy and definitive therapy showed no significant differences in microbiological success (Respectively, OR=6.00, 95% CI 0.42-85.25, OR=1.05 95% CI 0.17-6.52). There was also no statistically significant differences in the clinical success rate between the groups (OR=1.86, 95% CI 0.51-6.84, I=51%). CONCLUSIONS: There was no clear evidence that superiority of carbapenem and other alternative antibiotics. In patients urinary tract infection with ESBL-positive Enterobacteriaceae, the role of carbapenem and other antibiotics should be further evaluated for empirical and definitive treatment.