A Cochrane review included 28 studies with a total of 5 939 subjects. All trials were restricted to adults. The atypical arm consisted mostly of a quinolone (21 studies), or a macrolide (5 studies). The atypical antibiotic was administered as monotherapy in all but 3 studies. Only one study assessed a beta-lactam combined with a macrolide compared to the same beta-lactam. The non-atypical arm consisted of beta-lactams (BL) or cephalosporins. There was no difference in mortality between the atypical arm and the non-atypical arm (RR 1.14, 95% CI 0.84 to 1.55). The atypical arm showed an insignificant trend toward clinical success and a significant advantage to bacteriological eradication, which disappeared when evaluating methodologically high-quality studies alone. Clinical success for the atypical arm was significantly higher for Legionella pneumophilae (L. pneumophilae) and non-significantly lower for pneumococcal pneumonia. There was no significant difference between the groups in the frequency of (total) adverse events, or those requiring discontinuation of treatment. However, gastrointestinal events were less common in the atypical arm (RR 0.70, 95% CI 0.53 to 0.92).