Purpose: [18F]Fluoroethoxybenzovesamicol ([18F]FEOBV) is a radioligand for the vesicular acetylcholine transporter (VAChT), a marker of the cholinergic system. We evaluated the quantification of [18F]FEOBV in rats in control conditions and after partial saturation of VAChT using plasma and reference tissue input models and test-retest reliability. Procedure: Ninety-minute dynamic [18F]FEOBV PET scans with arterial blood sampling were performed in control rats and rats pretreated with 10 μg/kg FEOBV. Kinetic analyses were performed using one- (1TCM) and two-tissue compartmental models (2TCM), Logan and Patlak graphical analyses with metabolite-corrected plasma input, reference tissue Patlak with cerebellum as reference tissue, standard uptake value (SUV) and SUV ratio (SUVR) using 60- or 90-min acquisition. To assess test-retest reliability, two dynamic [18F]FEOBV scans were performed 1 week apart. Results: The 1TCM did not fit the data. Time-activity curves were more reliably estimated by the irreversible than the reversible 2TCM for 60 and 90 min as the influx rate Ki showed a lower coefficient of variation (COV, 14–24 %) than the volume of distribution VT (16–108 %). Patlak graphical analysis showed a good fit to the data for both acquisition times with a COV (12–27 %) comparable to the irreversible 2TCM. For 60 min, Logan analysis performed comparably to both irreversible models (COV 14–32 %) but showed lower sensitivity to VAChT saturation. Partial saturation of VAChT did not affect model selection when using plasma input. However, poor correlations were found between irreversible 2TCM and SUV and SUVR in partially saturated VAChT states. Test-retest reliability and intraclass correlation for SUV were good. Conclusion: [18F]FEOBV is best modeled using the irreversible 2TCM or Patlak graphical analysis. SUV should only be used if blood sampling is not possible.