UTIs are amongst the most frequent bacterial infections. However, the clinical phenotypes of UTI are heterogeneous and range from rather benign, uncomplicated infections to complicated UTIs (cUTIs), pyelonephritis and severe urosepsis. Stratification of patients with UTIs is, therefore, important. Several classification systems exist for the description and classification of UTIs, with the common rationale that cUTIs have a higher risk of recurrence or chronification, progression or severe outcome than uncomplicated UTIs. The pathophysiology and treatment of cUTIs and pyelonephritis are driven more by host factors than by pathogen attributes. cUTIs and pyelonephritis are associated with high antimicrobial resistance rates among causative pathogens. However, antimicrobial resistance rates can differ substantially, depending on the population being studied and whether the data being analysed are from surveillance studies, registry data or interventional studies, in which specific inclusion and exclusion criteria are used for patient selection. For example, antibiotic resistance rates are higher in patients with urosepsis than in those with less severe infections. Thus, treatment outcomes differ substantially among studies, ranging from 50% to almost 100% clearance of infection, depending on the patient population analysed, the UTI entities included and the primary outcome of the study. Pyelonephritis and cUTIs have emerged as infection models for the study of novel antibiotics, including extensive investigation of novel substances active against Gram-negative bacteria.