The reliability of the placement of regions of interest (ROIs) on emission computed tomograms (ECT) with a semi-automated magnetic resonance image (MRI)-guided method was compared with the reliability of manual ROI placement. MRI-defined regions were transferred to the registered, co-planar ECT image after accounting for inter-image differences in rotation, translation, and scaling. Six single photon emission computed tomographic (SPECT) scans of normal control subjects were rated for caudate perfusion using the radiotracer N-isopropyl[I-123]-iodoamphetamine (IMP). The MRI-guided method had higher interrater reliability, lower variance of region location, and lower variance of region activity than the manual method. The manual method recorded higher mean regional activity than the MRI-guided method, possibly due to sample bias inherent in the manual method.