BACKGROUND: The objective of this study was to assess temporal changes in incidence and short term mortality of Staphylococcus aureus bacteraemia (SAB) from 1995 through 2008. METHODS: The study was conducted as a nation-wide observational cohort study with matched population controls. The setting was hospitalized patients in Denmark 1995-2008. Uni- and multivariate analyses were used to analyze the hazard of death within 30 days from SAB. RESULTS: A total of 16 330 cases of SAB were identified: 57% were hospital-associated (HA), 31% were community-acquired (CA) and 13% were of undetermined acquisition. The overall adjusted incidence rate remained stable at 23 per 100 000 population but the proportion of SAB cases older than 75 years increased significantly. Comorbidity in the cohort as measured by Charlson comorbidity index (CCI) score and alcohol-related diagnoses increased over the study period. In contrast, among the population controls the CCI remained stable and alcohol-related diagnoses increased slightly. For HA SAB crude 30-day mortality decreased from 27.8% to 21.8% (22% reduction) whereas the change for CA SAB was small (26.5% to 25.8%). By multivariate Cox regression, age, female sex, time period, CCI score and alcohol-related diagnoses were associated with increased mortality regardless of mode of acquisition. CONCLUSIONS: Throughout a 14-year period the overall incidence of SAB remained stable while the overall short term prognosis continued to improve despite increased age and accumulation of comorbidity in the cohort. However, age and comorbidity were strong prognostic indicators for short term mortality.