The objective of this investigation was to assess whether between-hospital variation in echocardiography usage for patients with Staphylococcus aureus bacteraemia (SAB) is explained by differences in patients' pre-test probability of endocarditis. This was a retrospective cohort study at three neighbouring hospitals in Australia. Consecutive episodes of SAB were reviewed for the presence of three endocarditis risk factors (community onset, prolonged bacteraemia and the presence of an intracardiac prosthetic device) and the performance and results of all echocardiography studies within 30 days. Multivariate logistic regression was used to examine the effect of hospital site on the performance of (i) transoesophageal and (ii) transthoracic echocardiography controlling for major endocarditis risk factors. Significant variation in echocardiography usage was demonstrated between sites in a total cohort of 1167 episodes of SAB. None of the three sites were found to exhibit echocardiography usage that could be considered consistent with current guidelines, and each differed from the guidelines in different ways. Hospital site, rather than endocarditis risk factors, was the strongest predictor of transthoracic echocardiography use; however, the use of transoesophageal echocardiography was strongly predicted by endocarditis risk factors. Variation in echocardiography use between these hospitals is not adequately explained by differences in the risk factor profile of their SAB cohorts.