To prospectively compare the left ventricular end-diastolic volume (LVEDV), end-systolic volume (LVESV), and ejection fraction (LVEF) obtained from gated perfusion single photon emission computed tomography (GSPECT) with those obtained from cardiac magnetic resonance imaging (MRI) in patients with idiopathic dilated cardiomyopathy (IDC). Twenty-one patients with IDC (6 females) with a median age of 45 years (range 17-65) were scheduled for (99m)Tc-tetrofosmin-GSPECT and MRI within a 3 h interval. In both methods LV volumes were analyzed with the Simpson method. Both GSPECT and MRI were successfully completed in 90% of patients. Close linear correlations were observed between the two methods in LVEDV (R = 0.918; P < 0.001) and LVESV (R = 0.946; P < 0.001), but correlations were significantly weaker in LVEF (R = 0.323; P = 0.082). LVEDV and LVESV were smaller in GSPECT than in MRI (161 ± 73 vs. 214 ± 87 ml, P < 0.001, and 116 ± 64 vs. 138 ± 72 ml, P < 0.001, respectively). In 4 patients (21%) the LVEDV index was considered normal by GSPECT and increased by MRI, if MRI-derived normal values were used. No difference was found between GSPECT and MRI when LVEF(MRI) was ≤ 40%, but GSPECT showed smaller LVEF when LVEF(MRI) was over 40% (33 ± 11 vs. 50 ± 5%; P < 0.05). The finding of increased LVEDV in GSPECT is reliable compared with MRI. However, LV volumes were underestimated by GSPECT and no direct comparisons can be made between methods in follow up studies. Abnormal results should be confirmed by another imaging modality, such as MRI, if these findings have therapeutic consequences.