Abstract Background Recent guidelines suggest an extensive role of transesophageal echocardiography (TOE) in the diagnostic work-up for suspected infective endocarditis (IE). However, the negative predictive value of transthoracic echocardiography (TTE) remains poorly investigated. Purpose The purpose of our study was to evaluate the accuracy of TTE in ruling out IE in order to avoid unnecessary TOE, a more invasive and expensive diagnostic exam. Methods We reviewed data of 637 consecutive patients referred to our laboratory of echocardiography with a suspected diagnosis of IE who underwent TOE within 14 days of TTE (median 4.4 days; interquartile range 2.4-7.4). For each patient we obtained age, sex, blood culture results, complete blood count, blood chemistry, evidence of systemic embolism or pulmonary embolism, and presence of moderate/severe heart valve disease, congenital heart disease, valve prostheses and intracardiac devices. Patients with initial doubtful findings at TOE were referred for a follow-up TOE after one week and/or 18F fluorodeoxyglucose positron emission tomography computed tomography to achieve a definitive clinical diagnosis. Patients in whom a clear diagnosis of IE could not be achieved (n=7) were considered to have the disease. Results Of the 637 patients enrolled, 375 had negative TTE for IE (59%). TOE was positive and doubtful for IE in 44 (12%) and 29 (8%) of these patients, respectively. After further assessment, a final diagnosis of IE was achieved in 56 patients (15%). Variables associated with IE at univariable analysis included presence of valve prostheses (p<0.001), systemic or pulmonary embolism (p<0.001), lower platelet count (p=0.003), positive blood culture (p=0.005), aortic valve regurgitation (p=0.02), aortic valve stenosis (p=0.03) and mitral regurgitation (p=0.03). At multivariable logistic regression, independent predictors of IE (OR=odds ratio; CI=confidence interval) included evidence of systemic or pulmonary embolism (OR 17.75, 95% CI 6.66-47.3, p<0.001), bioprosthetic heart valves (OR 2.93, 95% CI 1.31-6.53 p=0.023) and lower platelet count (OR 0.995, 95% CI 0.991-0.998, p=0.003). Among the subgroup of patients (n=183, 49%) with no evidence of embolism, no bioprosthetic valve and platelet count >190,000/mL (median value), IE was diagnosed in 11 patients only (6%). Conclusions In our study, a negative TTE in the subgroup of patients with suspected IE showing platelet count >190,000/mL and absence of embolism and valvular bio-prosthesis, had a negative predictive value for IE of 94%, suggesting that, in this subset of patients, TOE can initially be delayed and performed only if clinical suspicion continues to remain high.