As part of our bone dating development, we have tested the ultrafiltration of bone gelatin using 2 different filters—Vivaspin 20™ (VS20), a polyethersulfone, and Vivaspin 15R™ (VS15R), a cellulose, both with a 30,000 molecular weight cutoff—and bone collagen from dated samples ranging in age from 1.5 to >50 kyr BP. A direct accelerator mass spectrometry (AMS) measurement yielded radiocarbon concentrations of ∼0.5 pMC (∼42 kyr) for the polyethersulfone, ∼14.4–17.5 pMC (∼15.6–14 kyr) for the cellulose, and ∼107.4 pMC for the glycerin. The filters were cleaned before use similar to the Oxford protocol (Bronk Ramsey et al. 2004), and a series of freeze-dried archaeological bone gelatin samples and a modern pig-skin gelatin were passed through VS20 and VS15R filters (Vivascience™). We recovered both the eluent (<30-kD fraction) and the liquid that stayed above the filter (>30 kD) in order to obtain a carbon mass and isotope balance. While the >30-kD collagen fraction that is usually selected for AMS analysis does not appear to be significantly contaminated, measurements show significant age differences between the eluent <30 kD and the unfiltered bone collagen, indicating that, despite cleaning, both glycerin and filter still give off contaminants in the eluent. Ultrafiltration with young collagen from pig skin generally confirms these results for the <30-kD fraction but also shows the possibility of small contaminations in the >30-kD fraction. Until a contamination with filter carbon of the >30-kD collagen fraction can be excluded, we would recommend caution in the use of ultrafiltration for cleaning bone collagen with VS20 or VS15R ultrafilters.