Acute otitis media (AOM) is one of the most common diseases in early infancy and childhood. Antibiotic use for AOM varies from 56% in the Netherlands to 95% in the USA, Canada and Australia. To assess the effects of antibiotics for children with AOM. We searched CENTRAL (2012, Issue 10), MEDLINE (1966 to October week 4, 2012), OLDMEDLINE (1958 to 1965), EMBASE (January 1990 to November 2012), Current Contents (1966 to November 2012), CINAHL (2008 to November 2012) and LILACS (2008 to November 2012). Randomised controlled trials (RCTs) comparing 1) antimicrobial drugs with placebo and 2) immediate antibiotic treatment with expectant observation (including delayed antibiotic prescribing) in children with AOM. Two review authors independently assessed trial quality and extracted data. For the review of antibiotics against placebo, 12 RCTs (3317 children and 3854 AOM episodes) from high-income countries were eligible. However, one trial did not report patient-relevant outcomes, leaving 11 trials with generally low risk of bias. Pain was not reduced by antibiotics at 24 hours (risk ratio (RR) 0.89; 95% confidence interval (CI) 0.78 to 1.01) but almost a third fewer had residual pain at two to three days (RR 0.70; 95% CI 0.57 to 0.86; number needed to treat for an additional beneficial outcome (NNTB) 20) and fewer had pain at four to seven days (RR 0.79; 95% CI 0.66 to 0.95; NNTB 20). When compared with placebo, antibiotics did not alter the number of abnormal tympanometry findings at either four to six weeks (RR 0.92; 95% CI 0.83 to 1.01) or at three months (RR 0.97; 95% CI 0.76 to 1.24), or the number of AOM recurrences (RR 0.93; 95% CI 0.78 to 1.10). However, antibiotic treatment did lead to a statistically significant reduction of tympanic membrane perforations (RR 0.37; 95% CI 0.18 to 0.76; NNTB 33) and halved contralateral AOM episodes (RR 0.49; 95% CI 0.25 to 0.95; NNTB 11) as compared with placebo. Severe complications were rare and did not differ between children treated with antibiotics and those treated with placebo. Adverse events (such as vomiting, diarrhoea or rash) occurred more often in children taking antibiotics (RR 1.34; 95% CI 1.16 to 1.55; number needed to treat for an additional harmful outcome (NNTH) 14). Funnel plots do not suggest publication bias. Individual patient data meta-analysis of a subset of included trials found antibiotics to be most beneficial in children aged less than two with bilateral AOM, or with both AOM and otorrhoea.For the review of immediate antibiotics against expectant observation, five trials (1149 children) were eligible. Four trials (1007 children) reported outcome data that could be used for this review. From these trials, data from 959 children could be extracted for the meta-analysis on pain at days three to seven. No difference in pain was detectable at three to seven days (RR 0.75; 95% CI 0.50 to 1.12). No serious complications occurred in either the antibiotic group or the expectant observation group. Additionally, no difference in tympanic membrane perforations and AOM recurrence was observed. Immediate antibiotic prescribing was associated with a substantial increased risk of vomiting, diarrhoea or rash as compared with expectant observation (RR 1.71; 95% CI 1.24 to 2.36). Antibiotic treatment led to a statistically significant reduction of children with AOM experiencing pain at two to seven days compared with placebo but since most children (82%) settle spontaneously, about 20 children must be treated to prevent one suffering from ear pain at two to seven days. Additionally, antibiotic treatment led to a statistically significant reduction of tympanic membrane perforations (NNTB 33) and contralateral AOM episodes (NNTB 11). These benefits must be weighed against the possible harms: for every 14 children treated with antibiotics, one child experienced an adverse event (such as vomiting, diarrhoea or rash) that would not have occurred if antibiotics had been withheld. Antibiotics appear to be most useful in children under two years of age with bilateral AOM, or with both AOM and otorrhoea. For most other children with mild disease, an expectant observational approach seems justified. We have no trials in populations with higher risks of complications.