BACKGROUND Community acquired pneumonia is responsible for a high morbidity in children. The etiological diagnosis is not always easy and treatment remains probabilistic. OBJECTIVE To analysis clinical patterns and the outcome of community acquired pneumonia and to find arguments in favour of its origin. METHODS A prospective analysis of the files of 39 children (20 boys, 19 girls) hospitalized between (1 december 2004 - 30 june 2005) for community acquired pneumonia was performed. Included dchildren whad between 6 months and 15 years-old. RESULTS The mean age was of 3 years and three months. Germ was identified in 41% of cases: Mycoplasma pneumoniae was the most important germ found in 9 cases, pneumococcus was found in 4 cases and hemophilus in four cases. Coinfection pneumococcus and mycoplasma were found in two cases. None virus was identified. We have separated two groups: bacterial pneumonia and pneumoniae without definite etiology. Fat cough was associated to bacterial pneumonia. mycoplama pneumoniae was significantly associated with high fever and dry cough. We haven't found any others associations between clinical, biological or radiological patterns and the two groups of pneumonia. C-reactive protein more than 66 mg/l has sensitivity of 77%, specificity of 73.3%, positive predictive value of 46.7% of and negative predictive value of 91.7%. The outcome was favourable in all cases. CONCLUSIONS Theses results showed the necessity to improve our microbiological methods to identify infectious agents of pneumonia. Mycoplasma pneumonia seems to be a frequent germ in preschool children.