OBJECTIVE: To determine the utility of 123I-metaiodobenzylguanidine cardiac scintigraphy (MIBG), and optimum heart: mediastinum ratio (HMR) for differentiating dementia with Lewy bodies (DLB) from Alzheimer's disease (AD) in a clinically representative population, comparing findings with those of 123I-2β -carbomethoxy-3β-(4-iodophenyl)-N-(3-fluoropropyl) nortropane (FP-CIT) SPECT. METHODS: We recruited subjects with probable DLB (n = 17) and probable AD (n = 16) from clinical services. Each participant underwent clinical examination, cardiac MIBG scintigraphy and FP-CIT SPECT. Diagnosis was made on the basis of clinical symptoms using validated criteria. Cardiac MIBG uptake was measured by the planar HMR, blind to clinical diagnosis, with values below a cut-off taken from a previous study (<2.2 at four hours) defining scans as abnormal. FP-CIT scans were blindly rated according to a visual rating scale. RESULTS: MIBG had a sensitivity, specificity and overall accuracy of 71%, 81% and 76% for distinguishing DLB from AD. FP-CIT demonstrated a sensitivity, specificity and accuracy of 82%, 88% and 85%. Using a lower HMR cut-off to distinguish between abnormal and normal MIBG scans improved the accuracy of MIBG, raising specificity (100%) and overall accuracy (85%) without compromising sensitivity (71%). Neither prescription of potentially interfering medications, nor a history of myocardial infarction (MI), had a significant effect on HMR. CONCLUSION: We found that MIBG did not demonstrate superior sensitivity and overall accuracy to FP-CIT. HMR cut-off influences biomarker utility, and clinical and Caucasian populations may require a lower cut-off than those reported elsewhere. Future MIBG studies should include clinically representative cohorts as neither medications nor previous MI appear to influence HMR.