BACKGROUND Isolating aetiological agents in patients with infective endocarditis (IE) remains problematical. We postulated that the high local incidence of culture-negative IE resulted from antibiotic exposure prior to blood cultures and that a structured delay in therapy in the subacute presentation would improve the diagnostic yield. OBJECTIVE We aimed to prospectively observe the diagnostic approach and give an overview of supplementary laboratory tests. METHODS Patients with suspected IE were enrolled into this analytical observational study and followed up for six months (n = 92). We compared the diagnostic yield and outcome in cases where antibiotics were withheld for 72 hours, with those cases who received early antimicrobials, despite being deemed safe for delayed therapy. RESULTS Definitive diagnoses (definite or excluded IE) were made in 92.8% of patients where antibiotics were delayed, compared to 60% of patients who received empirical treatment (p = 0.08). The mortality rates were 18.4% and 30.0% (p = 0.18). Twenty-three of 26 patients with definite culture-negative IE received antibiotics during the 48 hours preceding cultures, compared to eight of 21 culture-positive patients (P < 0.001). Screening for atypical bacteria did not improve the yield. C-reactive protein (CRP) had a sensitivity of 97.0% (negative predictive value 87.5%), whereas a positive rheumatoid factor (RF) had a specificity of 93.8% (positive predictive value 91.7%). CONCLUSIONS We observed tendencies towards a greater diagnostic yield and lower mortality where antibiotics were initially withheld. Antibiotic prior to blood cultures were an important cause of culture-negative IE. A normal CRP proved useful in excluding IE; a positive RF strongly favoured IE.