This study evaluates the reproducibility of the quantification of metabotropic glutamate receptor type 5 (mGluR5) densities in rats using the PET radiotracer [11C]ABP688 and pharmacokinetic models that are based on an input function, which is derived from a reference tissue. Seven rats underwent dynamic PET scans (60 min) after bolus injection of [11C]ABP688. Kinetic analyses included: binding potential (BPND) determined by calculating (a) the simplified reference tissue model (SRTM) and (b) its two-steps simplified version (SRTM2); (c) multilinear reference tissue model (MRTM) and (d) its 2-parameter version (MRTM2); (e) noninvasive graphical analysis (NIGA). Parametric images were generated representing BPND by the MRTM2 model. BPND determinations were reproducible with low to acceptable variability ranging from 5 to 10% and reproducibility scores (intraclass correlation coefficient) between 0.51 and 0.88. The pharmacokinetic model that showed lowest overall variability was the SRTM. In contrast, the use of the NIGA was associated with significantly lower reproducibility scores. Comparison of parametric images revealed no significant bias between test and retest measurements and is therefore suitable to compare groups at voxel levels. In conclusion, our results suggest that noninvasive quantification of [11C]ABP688 imaging is reproducible and reliable for PET studies of the cerebral mGluR5 in rats. Synapse, 2012. © 2012 Wiley Periodicals, Inc.