Although C-Reactive protein (CRP) was described more than 70 years ago and it still is commonly used in practice, studies exploring its usefulness persist while some propose its replacement by other inflammatory acute-phase-mediators. The aim of this clinical review is to answer the question if CRP measurement warrant clinical decisions for febrile children because it discriminates between bacterial from non bacterial etiologies. We made a systematic search by means of MEDLINE, SciELO and LILACS with the following MESH terms: "C-reactive protein", "bacterial", "infection", "children", "diagnosis" or "detection", besides the Haynes selector for articles on diagnosis, between 1950 and 2004. Selection data extraction and critical appraisal were independently made by the two authors, following standard criteria. We selected 7 primary articles, 3 clinical reviews and two randomized clinical trials. There was no disagreement between reviewers. Only one of the clinical reviews followed standard guidelines; two reviews concluded that no isolated CRP values would warrant to make decisions on starting or withholding antimicrobial therapy in febrile children. All primary articles showed methodological flaws in basic validity criteria. Both randomized clinical trials showed that CRP results did not change either doctor's decisions about antimicrobial prescriptions nor the studied patients' prognosis. We did not find evidences that could warrant the use of CRP for the defined problem in Pediatrics. Main problems affecting validity of studies on CRP as diagnostic resource are the lack of laboratory methods uniformity -including the gold standard make up- the heterogeneity of cut off points, clinical spectrum inappropriateness of study groups as well as the resulting diversity of the fixed indexes values. Moreover, no validation of this test in children population has been made so far