To assess the usefulness of in vivo 31P nuclear magnetic resonance (NMR) spectroscopy, we investigated spectra from the myocardium in 6 patients with old Q-wave infarction (QMI), 6 with old non-Q-wave infarction (NQMI) and 9 controls by ECG-gated depth-resolved surface-coil spectroscopy. External hexamethylphosphoric triamide (HMPT) was used to quantify the signal intensities. Left ventricular weight in the region of interest (LVW) was estimated from 1H magnetic resonance images. The extent score by 201Tl scintigraphy was determined in 3 QMI and 4 NQMI patients. No significant differences were found among the 3 groups in peak area ratios of 31P NMR spectra to phosphocreatine (PCr) or adenosine triphosphate (ATP). Compared with controls, significant reductions were observed in values for the peak areas of PCr normalized by the standard HMPT (PCr/HMPT) or by both HMPT and LVW (PCr/HMPT/LVW) for QMI patients (p less than 0.05), and in ATP/HMPT and ATP/HMPT/LVW for QMI and NQMI patients (p less than 0.01). There was a significant negative correlation between ATP/HMPT and the 201Tl scintigraphy extent score (p less than 0.05). These findings suggest that in vivo 31P NMR spectroscopy can detect high-energy phosphate reduction in the infarcted myocardium and may be useful in evaluating myocardial viability.