Atypical bacteria are treatable causative agents of community-acquired pneumonia (CAP). However, there is no conclusive evidence that a child with CAP should receive empirical treatment against such agents. Objectives: We assessed the possibility of association between clinical failure and acute infection by these bacteria among children with CAP treated with amoxicillin. Patients and methods: Patients aged 2–59months with non-severe CAP received amoxicillin during prospective follow-up. Acute and convalescent blood samples were collected. Probable acute infection by Mycoplasma pneumoniae (specific IgM antibodies), by Chlamydia pneumoniae or Chlamydia trachomatis (specific IgM antibodies and/or IgG/IgA titre change) was investigated. Outcomes were assessed during follow-up at 2, 5 and 14–28 days. Treatment failure included development of danger signs, persistent fever, tachypnoea or death. ClinicalTrials.gov: NCT01200706. Results: Of 787 children, 86 (10.9%; 95% CI"8.9%–13.3%) had acute M. pneumoniae infection. C. pneumoniae acute infection was found in 79 of 733 (10.8%; 95% CI"8.7%–13.2%) and C. trachomatis was found in 3 of 28 (10.7%; 95%CI"2.8%–26.5%) ,6months old. Among patients with or without treatment failure at 2 days, acute M. pneumoniae infection (11.7% versus 10.7%; P"0.7), acute C. pneumoniae infection (8.5% versus 11.3%; P"0.3) and acute C. trachomatis infection (16.7% versus 9.1%; P"0.5) were found. No significant differences were found with regard to treatment failure at the 5 day evaluation. Overall, amoxicillin was substituted in 3.5% versus 2.7%among patientswith orwithout acute infection by one of these bacteria (P"0.6). Conclusions: The overall substitution rate of amoxicillin was very low. It is not necessary to give an empirical nonb- lactamantibiotic as a first-line option to treat every child between 2 and 59months old with non-severe CAP.