Clostridium difficile (C. difficile) is recognized as a frequent cause of antibiotic-associated diarrhoea and colitis. This review is an update of a previously published Cochrane review. The aim of this review is to investigate the efficacy and safety of antibiotic therapy for C. difficile-associated diarrhoea (CDAD), or C. difficile infection (CDI), being synonymous terms. We searched MEDLINE, EMBASE, CENTRAL and the Cochrane IBD Group Specialized Trials Register from inception to 26 January 2017. We also searched clinicaltrials.gov and clinicaltrialsregister.eu for ongoing trials. Only randomised controlled trials assessing antibiotic treatment for CDI were included in the review. Three authors independently assessed abstracts and full text articles for inclusion and extracted data. The risk of bias was independently rated by two authors. For dichotomous outcomes, we calculated the risk ratio (RR) and corresponding 95% confidence interval (95% CI). We pooled data using a fixed-effect model, except where significant heterogeneity was detected, at which time a random-effects model was used. The following outcomes were sought: sustained symptomatic cure (defined as initial symptomatic response and no recurrence of CDI), sustained bacteriologic cure, adverse reactions to the intervention, death and cost. Twenty-two studies (3215 participants) were included. The majority of studies enrolled patients with mild to moderate CDI who could tolerate oral antibiotics. Sixteen of the included studies excluded patients with severe CDI and few patients with severe CDI were included in the other six studies. Twelve different antibiotics were investigated: vancomycin, metronidazole, fusidic acid, nitazoxanide, teicoplanin, rifampin, rifaximin, bacitracin, cadazolid, LFF517, surotomycin and fidaxomicin. Most of the studies were active comparator studies comparing vancomycin with other antibiotics. One small study compared vancomycin to placebo. There were no other studies that compared antibiotic treatment to a placebo or a 'no treatment' control group. The risk of bias was rated as high for 17 of 22 included studies. Vancomycin was found to be more effective than metronidazole for achieving symptomatic cure. Seventy-two per cent (318/444) of metronidazole patients achieved symptomatic cure compared to 79% (339/428) of vancomycin patients (RR 0.90, 95% CI 0.84 to 0.97; moderate quality evidence). Fidaxomicin was found to be more effective than vancomycin for achieving symptomatic cure. Seventy-one per cent (407/572) of fidaxomicin patients achieved symptomatic cure compared to 61% (361/592) of vancomycin patients (RR 1.17, 95% CI 1.04 to 1.31; moderate quality evidence). Teicoplanin may be more effective than vancomycin for achieving a symptomatic cure. Eightly-seven per cent (48/55) of teicoplanin patients achieved symptomatic cure compared to 73% (40/55) of vancomycin patients (RR 1.21, 95% CI 1.00 to 1.46; very low quality evidence). For other comparisons including the one placebo-controlled study the quality of evidence was low or very low due to imprecision and in many cases high risk of bias because of attrition and lack of blinding. One hundred and forty deaths were reported in the studies, all of which were attributed by study authors to the co-morbidities of the participants that lead to acquiring CDI. Although many other adverse events were reported during therapy, these were attributed to the participants' co-morbidities. The only adverse events directly attributed to study medication were rare nausea and transient elevation of liver enzymes. Recent cost data (July 2016) for a 10 day course of treatment shows that metronidazole 500 mg is the least expensive antibiotic with a cost of USD 13 (Health Warehouse). Vancomycin 125 mg costs USD 1779 (Walgreens for 56 tablets) compared to fidaxomicin 200 mg at USD 3453.83 or more (Optimer Pharmaceuticals) and teicoplanin at approximately USD 83.67 (GBP 71.40, British National Formulary). No firm conclusions can be drawn regarding the efficacy of antibiotic treatment in severe CDI as most studies excluded patients with severe disease. The lack of any 'no treatment' control studies does not allow for any conclusions regarding the need for antibiotic treatment in patients with mild CDI beyond withdrawal of the initiating antibiotic. Nonetheless, moderate quality evidence suggests that vancomycin is superior to metronidazole and fidaxomicin is superior to vancomycin. The differences in effectiveness between these antibiotics were not too large and the advantage of metronidazole is its far lower cost compared to the other two antibiotics. The quality of evidence for teicoplanin is very low. Adequately powered studies are needed to determine if teicoplanin performs as well as the other antibiotics. A trial comparing the two cheapest antibiotics, metronidazole and teicoplanin, would be of interest.