OBJECTIVE: The aim of our study was to show that spatial resolution can be improved without loss of diagnostic accuracy if a 3D inversion recovery gradient-recalled echo (GRE) sequence is used instead of a segmented inversion recovery GRE at 3 T for the assessment of myocardial infarction. SUBJECTS AND METHODS: Fifteen patients with myocardial infarction were examined on a 3-T MR system. A segmented breath-hold 3D inversion recovery GRE technique with a voxel size of 6.3 mm(3) was compared with a breath-hold standard 2D inversion recovery GRE technique with a voxel size of 21.3 mm(3) for the detection of delayed enhancement. Contrast-to-noise ratios (CNRs) were calculated and infarct volumes were measured. Detection and transmural extent of infarctions were evaluated using kappa statistics. Total acquisition times were measured for both sequences. RESULTS: The CNR in the 3D technique did not show any significant difference compared with the 2D technique. The correlation coefficients of the infarct volumes determined with the 3D and 2D inversion recovery GRE studies at 3 T were r = 0.99 (p< 0.001). The assessment of the presence of hyperenhanced myocardium in all segments and the evaluation of transmurality resulted in very good agreement (kappa = 0.98 and kappa = 0.90). Total acquisition time was significantly shorter with the 3D technique (2.4 +/- 0.9 minutes) than with the 2D technique (4.9 +/- 1.5 minutes) (p< 0.001). CONCLUSION: The use of a 3D inversion recovery GRE sequence at 3 T allows accurate assessment of myocardial infarction without loss of CNR compared with the standard 2D technique. Furthermore, data acquisition time can be significantly reduced.