In an ongoing international multi-centre trial, positron emission tomography (PET) is being used to evaluate the effect of a new P-selectin antagonist on the infarct size in patients with acute myocardial infarction, treated with thrombolysis. Although it is possible to correct for site-dependent factors, it is desirable to reduce these factors to a minimum. Therefore, acquisition and reconstruction protocols have been defined that can be closely followed by all participating centres. The resulting reconstructed images are transferred to the core centre for central processing with semi-automatic software. This paper reports on the multi-centre phantom experiment that was carried out to assess the inter-centre reproducibility of defect size determination with this protocol. Also, the spatial resolution of the short axis slices was examined. In addition, the analysis procedure was applied to normal PET studies to evaluate the specificity of perfusion defect detection. The transmural cold defect in the phantom occupied 14.8% of the left ventricular area. The automated analysis was applied to the phantom measurements from the 14 participating PET cameras. It yielded an accurate estimate of 15.1% with a standard deviation of 0.6%, indicating excellent reproducibility. The spatial resolution in the short axis slices was similar for all PET systems: 9.6+/-0.8 mm. The same procedure produced a defect size of zero in the studies of normal volunteers. This study indicates that cardiac studies from multiple PET systems can be pooled for statistical analysis.