BACKGROUND: In order to determine myocardial salvage, accurate quantification of myocardium at risk (MaR) is necessary. We present a validated novel automatic segmentation algorithm for quantification of MaR by myocardial perfusion SPECT (MPS) in patients with acute coronary occlusion. METHODS AND RESULTS: Twenty-nine patients with coronary occlusion were injected with a perfusion tracer before reperfusion, and underwent rest MPS within 4 hours. The MaR was quantified using the proposed algorithm (Segment software), the software Quantitative Perfusion SPECT (QPS) and by manual segmentation. The Segment MaR algorithm used a threshold of 55% of maximal counts and an a priori model based on normal coronary artery perfusion territories. The MaR was 30 +/- 10% left ventricular mass (%LVM) by manual segmentation, 31 +/- 12%LVM by Segment, and 36 +/- 14%LVM by QPS. There was a good agreement between automatic and manual segmentation for both of the algorithms with a lower bias for Segment (.8 +/- 4.0%LVM) than for QPS (5.8 +/- 5.8%LVM) when compared to manual segmentation. CONCLUSIONS: The Segment MaR algorithm can be used to correctly assess MaR from MPS images in patients with acute coronary occlusion without access to tracer-specific normal database. The MaR in relation to final infarct size enables determination of myocardial salvage.