OBJECTIVES: To study between-hospital variation in mortality, readmissions and prolonged length of stay across Belgian hospitals. DESIGN: A retrospective nationwide observational study. SETTING: Secondary and tertiary acute-care hospitals in Belgium. PARTICIPANTS: We studied 4 560 993 hospital stays in 99 (98%) Belgian acute-care hospitals between 2016 and 2018. PRIMARY OUTCOME MEASURES: Using generalised linear mixed models, we calculated hospital-specific and Major Diagnostic Category (MDC)-specific risk-adjusted in-hospital mortality, readmissions within 30 days and length of stay above the MDC-specific 90th percentile and assessed between-hospital variation through estimated variance components. RESULTS: There was strong evidence of between-hospital variation in mortality, readmissions and prolonged length of stay across the vast majority of patient service lines. Overall, should hospitals with upper-quartile risk-standardised rates succeed in improving to the median level, a yearly 4076 hospital deaths, 3671 readmissions and 15 787 long patient stays could potentially be avoided in those hospitals. Our analysis revealed a select set of 'high-impact-opportunity hospitals' characterised by poor performance across outcomes and across a large number of MDCs. CONCLUSIONS: Analysis of between-hospital variation highlights important differences in patient outcomes that are not explained by known patient or hospital characteristics. Identifying 'high-impact-opportunity hospitals' can help government inspection bodies and hospital managers to establish targeted audits and inspections to generate effective quality improvement initiatives.