BACKGROUND: Use of antivirals is recommended for the control of seasonal and pandemic influenza. Our aim was to review the evidence of efficacy, effectiveness, and safety of registered antivirals against naturally occurring influenza in healthy adults. METHODS: We searched various Databases to October, 2005, and contacted manufacturers and corresponding authors. We included randomised controlled trials comparing prophylactic (n=27) or treatment (n=27) efficacy against symptomatic or asymptomatic influenza. We did a meta-analysis and expressed prophylactic efficacy as a proportion (1-relative risk [RR]). For treatment trials, because of inconsistent and non-standardised reporting, we expressed continuous outcomes either as means or as hazard ratios. FINDINGS: We included 51 reports of 52 randomised controlled trials. Amantadine prevented 61% (95% CI 35-76) of influenza A cases and 25% (13-36) of cases of influenza-like illness, but caused nausea (OR 2.56, 1.37-4.79), insomnia and hallucinations (2.54, 1.50-4.31), and withdrawals because of adverse events (2.54, 1.60-4.06). There was no effect on asymptomatic cases (RR 0.85, 0.40-1.80). In treatment, amantadine significantly shortened duration of fever compared with placebo (by 0.99 days, -1.26 to -0.71), but had no effect on nasal shedding of influenza A viruses (0.93, 0.71-1.21). The fewer data for rimantadine showed comparable effects. In prophylaxis, compared with placebo, neuraminidase inhibitors have no effect against influenza-like illness (1.28, 0.45-3.66 for oral oseltamivir 75 mg daily, 1.51, 0.77-2.95 for inhaled zanamivir 10 mg daily). Higher doses appear to make no difference. The efficacy of oral oseltamivir 75 mg daily against symptomatic influenza is 61% (15-82), or 73% (33-89) at 150 mg daily. Inhaled zanamivir 10 mg daily is 62% efficacious (15-83). Neither neuraminidase inhibitor appeared effective against asymptomatic influenza. Oseltamivir induces nausea (OR 1.79, 1.10-2.93), especially at higher prophylactic doses (2.29, 1.34-3.92). Oseltamivir in a post-exposure prophylaxis role has a protective efficacy of 58.5% (15.6-79.6) for households and from 68% (34.9-84.2) to 89% (67-97) in contacts of index cases. In influenza cases, compared with placebo the hazard ratios for time to alleviation of symptoms were 1.33, 1.29-1.37 for zanamivir; 1.30, 1.13-1.50 for oseltamivir provided medication was started within 48 h of symptom onset. Viral nasal titres were significantly diminished by both drugs (weighted mean difference -0.62, -0.82 to -0.41). Oseltamivir at 150 mg daily was effective in preventing lower respiratory tract complications in influenza cases (OR 0.32, 0.18-0.57). We could find no credible data on the effects of oseltamivir on avian influenza. INTERPRETATION: The use of amantadine and rimantadine should be discouraged. Because of their low effectiveness, neuraminidase inhibitors should not be used in seasonal influenza control and should only be used in a serious epidemic or pandemic alongside other public-health measures.