Resistance is one of failure's reasons. We tried, through clinical experience, to approach the magnitude and nature of the links, between phenotypically defined acquired resistance and clinical failure, in community acquired respiratory infections. An efficient resistance mecanism, able to suppress antibiotic action, is clearely associated to a risk of clinical failure (e.g. betalactamase secretion, target modification using methilation for macrolides, target mutation for fluoroquinoles). Resistance mecanism due to reduction of target affinity (pneumococcus and betalactams) progressively decreasing beta lactam activity depending on its expression, is at present time, not clearely associeted with clinical failure. Critical concentration, defining phenotypical resistance, is predictive of failure if it identifies a bacterial population owning an efficient resistance mecanism. It will not be predictive of failure if that concentration do not detect the resistance mecanism (e.g. parC mutation and levofloxacin) or if the link between antibiotic and resistant bacteria is not binary but depends also on pharmacokinetic parameters (pneumococcus and betalactam). Using resistance as a parametre for antibiotic choice, must integrate several elements: presence or not of a resistance mecanism, type and efficiency of the mecanism, links with clinical failure and antibiotic concentration, type and site of infection. Critical concentration is not allways the magic number that predict failure or success.