Four hundred and forty-one and 1048 episodes of bacteraemia were prospectively surveyed over a period of 18 months in two hospitals, a 450 bed community hospital and a 900 bed tertiary care urban university hospital. Incidence of bacteraemia was 2.18 per 1000 hospitalization days (10.1 per 1000 admissions) in the community hospital and 2.64 per 1000 hospitalization days (12.0 per 1000 admissions (P < 0.004)) in the university hospital. Sixty six and 62% of episodes of bacteraemia were community acquired. The majority of bacteraemic episodes originated on the internal medicine wards of both hospital—46.7% and 58.7% respectively; the incidence of bacteraemia in the medical divisions of both hospitals was 23.1 and 17.5 per 1000 admissions respectively (PEscherichia coli was the commonest Gram-negative pathogen in both hospitals, particularly the community hospital—47.5% vs 32.8% (P < 0.005) of all Gram-negative pathogens, while Pseudomonas spp. were significantly more common in the university hospital—18.5% vs 11.8% (P < 0.02). Non-enterococcal streptococci were more common in the community hospital while enterococci were far more common at the university hospital—15.1% vs 1% of all Gram-positive pathogens (P < 0.05). Staphylococcus epidermidis was more common among the community hospital Gram-positive bacteraemias—31.1% vs 18.6% (P < 0.005). For almost all genera and species, antibiotic resistance was higher at the university hospital. Twenty nine point four per cent of Staphylococcus aureus isolates from the university hospital were methicillin resistant compared to 2.4% at the community hospital (P < 0.005). 29.4% of all Streptococcus pneumoniae isolates at the university hospital were penicillin resistant while no resistance was found at the community hospital. A high resistance rate to ofloxacin was found at the university hospital among S. aureus and Pseudomonas sp. Sources of bacteraemia did not differ significantly between the two hospitals. In conclusion, although outcome did not differ significantly for the two hospitals, there were significant differences between blood culture isolates in these two different settings. These differences may influence clinical decision-making about antibiotic therapy for patients in these hospitals.