The objective was to compare the epidemiology and outcome of healthcare- (HA-) and community-associated (CA-) MRSA bacteraemia. A 10-year retrospective study of MRSA bacteraemia was carried out. Episodes were classified according to established criteria. Molecular typing was performed on a subset of isolates. Of 197 MRSA bacteraemia episodes, 178 (90.4%) were classified as HA-MRSA and 19 (9.6%) as CA-MRSA. All-cause 7- and 30-day mortality rates were similar in the HA and CA-MRSA bacteraemia groups; however, 1-year mortality was higher in the HA-MRSA bacteraemia group (48.3% vs 21.1% [p = 0.023]). Thirty-day all-cause mortality was significantly lower if empiric antimicrobial therapy included agent(s) to which the isolate tested susceptible, compared with patients receiving “inactive” therapy (19% vs 35.1% [p = 0.011]). The majority of MRSA bacteraemia episodes were caused by clones known to circulate in the community. All-cause mortality is as high in HA- as in CA-MRSA bacteraemia. Thirty-day mortality was significantly reduced if the patient received an antibiotic with activity against the MRSA isolate.