Lower respiratory tract infection is one of the leading killers of children worldwide, and developing countries bear the largest burden of mortality and morbidity from this disease. This Cochrane systematic review analyses the effect of several antibiotics or combinations of antibiotics in community acquired pneumonia (CAP) in children under 18 years of age, including only randomized controlled trials that used the WHO case definition of pneumonia or radiologically confirmed disease. Studies of hospital-treated and ambulatory cases were included, and the outcome of treatment with different antibiotics was compared The review analysed 20 studies that met the inclusion criteria with 22 antibiotic comparisons. However, only four comparisons of the same antibiotics were investigated in more than one study. Most studies did not find evidence of a difference in any of the outcomes that were analysed. The only positive relevant findings were: in ambulatory patients a higher failure rate of co-trimoxazole compared with amoxicillin (OR 1.33; 95% CI 1.05-1.67), higher cure rates of procaine penicillin compared with co-trimoxazole (OR 2.64; 95% CI 1.57-4.45), and higher cure rates of coamoxyclavulanic acid compared with amoxicillin (OR 10.44; 95% CI 2.85-38.21); in hospitalized patients higher readmission rates with chloramphenicol compared to the penicillin-gentamicin combination (OR 1.6; 95% CI 1.02-2.55). All the other studied comparisons were not significantly different.Knowing the expected pathogens is vital in choosing empiric antibiotics for pneumonia. In ten of the 20 studies, identification of bacterial etiologic agents was attempted with blood cultures with or without serology. Bacterial pathogens could be identified in *Correspondence to: Dr Melchor Sánchez-Mendiola, Medical Education Research Unit, Postgraduate Studies Division, National Autonomous University of Mexico Faculty of Medicine, Mexico City, Mexico. E-mail: melchors@liceaga.facmed.unam.mx only 8% of the cases (52% had Streptococcus pneumoniae, 32% Haemophilus influenzae, 3% Staphylococcus aureus and 13% other pathogens). One paper in the review explicitly included malnourished patients, and there is no information provided about the local immunization practices against S. pneumoniae and H. influenzae type B during the study period. These two etiologic agents constituted 84% of all bacterial isolates. This information is important since immunization practices modify the epidemiology of the disease and countries that have implemented effective immunization programs incorporate this information in the diagnostic and therapeutic decision-making process of physicians (1). Furthermore, the review does not mention the countries of origin of the studies and does not clarify if all occurred in developing countries, which would help the reader to judge the expected etiologic agents. Although studies of children with definite immunodeficiency were excluded, it seems likely that children with undiagnosed human immunodeficiency virus could have been included in so...