BACKGROUND: Primary-care physicians continue to overprescribe antibiotics for acute rhinosinusitis because distinction between viral and bacterial sinus infection is difficult. We undertook a meta-analysis of randomised trials based on individual patients' data to assess whether common signs and symptoms can be used to identify a subgroup of patients who benefit from antibiotics. METHODS: We identified suitable trials--in which adult patients with rhinosinusitis-like complaints were randomly assigned to treatment with an antibiotic or a placebo--by searching the Cochrane Central Register of Controlled Trials, Medline, and Embase, and reference lists of reports describing such trials. Individual patients' data from 2547 adults in nine trials were checked and re-analysed. We assessed the overall effect of antibiotic treatment and the prognostic value of common signs and symptoms by the number needed to treat (NNT) with antibiotics to cure one additional patient. FINDINGS: 15 patients with rhinosinusitis-like complaints would have to be given antibiotics before an additional patient was cured (95% CI NNT[benefit] 7 to NNT[harm] 190). Patients with purulent discharge in the pharynx took longer to cure than those without this sign; the NNT was 8 patients with this sign before one additional patient was cured (95% CI NNT[benefit] 4 to NNT[harm] 47). Patients who were older, reported symptoms for longer, or reported more severe symptoms also took longer to cure but were no more likely to benefit from antibiotics than other patients. INTERPRETATION: Common clinical signs and symptoms cannot identify patients with rhinosinusitis for whom treatment is clearly justified. Antibiotics are not justified even if a patient reports symptoms for longer than 7-10 days.