For serial imaging studies of the rat brain with positron emission tomography (PET), reproducible positioning of the head can facilitate spatial alignment of images and quantitative analysis. To achieve this aim, we constructed a plastic head frame and tested the positioning reproducibility on a high-resolution small-animal PET scanner, microPET. Two sets of ear bars, with tapers of either 18 degrees (sharp) or 45 degrees (blunt), were evaluated for their relative precision in securing the animal to the frame. For sequential positioning of an animal, average distances from the mean position of 0.51 mm (SD 0.41 mm) and 0.91 mm (SD 0.48 mm) were measured with the sharp and blunt ear bars, respectively. These results show that a rat brain can be reproducibly positioned using the frame, with a variation of position less than the spatial resolution of modern animal PET scanners. Brain regions of interest defined on one scan and copied across subsequent scans of a frame-repositioned animal resulted in an average coefficient of variation of 5.4% (SD 2.7%) using the sharp ear bars and 6.8% (SD 2.5%) using the blunt ear bars. This methodology has the potential to improve quantitative assessment for serial PET studies.