The last decade has seen a significant improvement in published evidence to show the clinical predictive value of phenotypic susceptibility testing with categorization of pathogens as 'susceptible' or 'resistant' based on clinical breakpoints. Most of the published data are based on retrospective or prospective observational clinical studies of patients treated with appropriate [pathogen(s)-susceptible] or inappropriate [pathogen(s)-resistant] chemotherapy. Appropriate therapy has been shown to improve outcomes in infections occurring in hospitals, such as bloodstream infection (BSI) and pneumonia in the intensive care unit. Infections due to specific pathogens such as extended-spectrum beta-lactamase-producing Enterobacteriaceae, Pseudomonas aeruginosa and Staphylococcus aureus also respond better to appropriate than inappropriate antibiotics. The situation with vancomycin-resistant enterococci is less clear, perhaps due to the increased importance of patient confounders. Streptococcus pneumoniae when causing acute pneumonia with or without BSI is a well-known exception to the predictive value of laboratory-defined resistance. Antibiotic resistance also impacts on outcomes in the community where the evidence is best for urinary tract infection. The clinical studies are compatible with the current pharmacokinetic/pharmacodynamic paradigm used to explain and predict antibacterial effects and therefore have a sound basis in antimicrobial science. These data underline the importance of well-constructed epidemiological studies to determine the prevalence of antimicrobial resistance in clinical practice and the central place of laboratory-based susceptibility testing in dictating antimicrobial therapy and so optimizing patient outcomes.