BACKGROUND Epidemiological evidence has suggested a link between beta-agonists and increases in asthma mortality. There has been much debate about possible causal links for this association, and whether regular (daily) long-acting beta(2)-agonists are safe. OBJECTIVE The aim of this review is to assess the risk of fatal and non-fatal serious adverse events in trials that randomised patients with chronic asthma to regular salmeterol versus placebo or regular short-acting beta(2)-agonists. METHODS Trials were identified using the Cochrane Airways Group Specialised Register of trials. Web sites of clinical trial registers were checked for unpublished trial data and FDA submissions in relation to salmeterol were also checked. The date of the most recent search was October 2007. METHODS Controlled parallel design clinical trials on patients of any age and severity of asthma were included if they randomised patients to treatment with regular salmeterol and were of at least 12 weeks duration. Concomitant use of inhaled corticosteroids was allowed, as long as this was not part of the randomised treatment regimen. METHODS Two authors independently selected trials for inclusion in the review. Outcome data was extracted by one author and checked by the second author. Unpublished data on mortality and serious adverse events was sought. RESULTS The review includes 26 trials comparing salmeterol to placebo and 8 trials comparing with salbutamol. These included 62,630 participants with asthma (including 2,380 children). In 6 trials (2,766 patients), no serious adverse event data could be obtained. All cause mortality was higher with regular salmeterol than placebo but the increase was not significant, Odds Ratio 1.33 [95% CI: 0.85, 2.10]. Non-fatal serious adverse events were significantly increased when regular salmeterol was compared with placebo, Odds Ratio 1.14 [95% CI: 1.01, 1.28]. One extra serious adverse event occurred over 28 weeks for every 188 people treated with regular salmeterol [95% CI: 95 to 2606]. There is insufficient evidence to assess whether the risk in children is higher or lower than in adults. No significant increase in fatal or non-fatal serious adverse events was found when regular salmeterol was compared with regular salbutamol. Individual patient data from the SNS study have been combined with the results of the SMART study; in patients who were not taking inhaled corticosteroids, compared to regular salbutamol or placebo, there was a significant increase in risk of asthma-related death with regular salmeterol, Odds Ratio 9.52 [95% CI: 1.24, 73.09]. The confidence interval for patients taking inhaled corticosteroids is too wide to rule out an increase in asthma mortality in this group. CONCLUSIONS In comparison with placebo, we have found an increased risk of serious adverse events with regular salmeterol. There is also a clear increase in risk of asthma-related mortality in patients not using inhaled corticosteroids in the two large surveillance studies. Although the increase in asthma-related mortality was smaller in patients taking inhaled corticosteroids at baseline, the confidence interval is wide, so it cannot be concluded that the inhaled corticosteroids abolish the risks of regular salmeterol. The adverse effects of regular salmeterol in children remain uncertain due to the small number of children studied.