Objectives To assess benefits and harms of empirical mono- vs. combination antibiotic therapy in adult patients with severe sepsis in the intensive care unit (ICU). Methods We performed a systematic review according to the Cochrane Collaboration methodology, including meta-analysis, risk of bias assessment and trial sequential analysis (TSA). We included randomised clinical trials (RCT) assessing empirical mono-antibiotic therapy versus a combination of two or more antibiotics in adult ICU patients with severe sepsis. We exclusively assessed patient-important outcomes, including mortality. Two reviewers independently evaluated studies for inclusion, extracted data, and assessed risk of bias. Risk ratios (RRs) with 95% confidence intervals (CIs) were estimated and the risk of random errors was assessed by TSA. Results Thirteen RCTs (n = 2633) were included; all were judged as having high risk of bias. Carbapenems were the most frequently used mono-antibiotic (8 of 13 trials). There was no difference in mortality (RR 1.11, 95% CI 0.95–1.29; p = 0.19) or in any other patient-important outcomes between mono- vs. combination therapy. In TSA of mortality, the Z-curve reached the futility area, indicating that a 20% relative risk difference in mortality may be excluded between the two groups. For the other outcomes, TSA indicated lack of data and high risk of random errors. Conclusions This systematic review of RCTs with meta-analysis and TSA demonstrated no differences in mortality or other patient-important outcomes between empirical mono- vs. combination antibiotic therapy in adult ICU patients with severe sepsis. The quantity and quality of data was low without firm evidence for benefit or harm of combination therapy.