Infective endocarditis (IE) is frequently suspected but infrequently diagnosed disease in children. Clinicians often order echcardiograms to rule out IE. In the era of cost containment, clinically efficient strategies must be developed. We hypothesized that transthoracic echocardiography (TTE) is only useful in children in whom there is a high clinical suspicion of IE based on history, physical examination and persistently positive blood cultures. The objective of this study is to determine the role of TTE as a screening test for suspected IE in children. Echocardiographic reports and medical records were reviewed for 166 consecutive patients who underwent TTE to rule out IE from 1/93 to 7/96. Fever was the main indication for TTE (114 patients, 69%). Fifty seven (34%) patients had congenital heart disease and 89 (54%) patients had indwelling venous catheters. Twenty six (16%) were diagnosed and treated as IE. TTE was positive for vegetation in 12 patients (46% of IE patients and 7% of the total). The other 14 patients were diagnosed on a clinical basis as well as having persistently positive blood cultures. Using logistic regression and classification trees, we determined which variables were predictive of IE. Fever, leukocytosis, splenomegaly, hematuria, heart failure, murmurs, immune compromised status, or elevated acute phase reactants were not predictive of the outcome. Malaise defined as fatigue and the general feeling of illness, was the most predictive variable of IE (sensitivity 77% and specificity 95%). The presence of indwelling catheters without malaise was indicative of no IE in 77/77 patients. The presence of a second positive blood culture or malaise increased the sensitivity for IE to 100% with only 50(instead of 166) patients requiring TTE under these criteria. TTE has poor sensitivity as a screening test for IE. Multiple positive blood cultures in a setting of clinical illness adequately defines the need for TTE.