Empiric antibiotic therapy in emergency remains frequent in community-acquired pneumonia in children primarily because of the high number of different causes. Streptococcus pneumoniae results in severe pneumonia and represents between 15 to 30% of etiologies. Lack of specificity of diagnostic procedures is important. Lobar consolidation is radiologically seen in less than half of cases and laboratory data, except for high procalcitonin level, are poorly reliable. Pneumonia due to Mycoplasma pneumoniae are frequent after age of 2 years, reaching 40 to 60% of causes in ambulatory teenagers. They must be given macrolides without important delay because sequellae are possible. The exact number of viral pneumonia is difficult to establish because of lack of reliable diagnostic methods. Bacterial superinfections are probably overestimated during acute phase but viral infections may lead to bacterial pneumonia 2 to 4 weeks after the initial episode. In absence of specific clinical or laboratory data, empiric antibiotic treatment must include pneumococci and their penicillin-resistant strains. Amoxicillin is the antibiotic of choice with a higher efficacy on resistant pneumococci than oral cephalosporins. In case of clinical failure of amoxicillin, mycoplasma infection is highly probable and patient must receive macrolides. Epidemiology is progressively changing with anti-pneumococcal immunisation but difficulties in diagnosis and in choice of empiric antibiotic treatment will remain important. Future studies in immunised children are needed to check the importance of pneumococcal infections due to serotypes not included in the vaccines