Wear of the polyethylene tibial component is a major factor in the success of total knee replacements. However, sampling resolution and the challenges of estimating original surfaces with relatively complex articulating geometries have limited the accuracy of volumetric measurements of wear on surgically retrieved inserts. A mathematical model analyzed volume error due to sampling resolution and found that 100 × 100 μm2 point spacing reduced error below 1 mm3. Small volumes of material were progressively removed from the topside of three unworn tibial inserts, after which each component was weighed and digitized with a laser coordinate measuring machine. Six inserts worn in knee simulator tests and nine surgically retrieved inserts visually scored for damage were also digitized. For these tests, the original surface of an insert was mathematically reconstructed from unworn regions of the same component, and volume loss and its spatial distribution were calculated. Volume loss estimated by autonomous reconstruction correlated strongly to mass removed manually (R2 = 0.954, slope = 1.02 ± 0.04), mass lost during simulator testing (R2 = 0.935, slope = 1.01 ± 0.07) and visual damage scores separated by size (R2large = 0.9824, R2small = 0.9728). These results suggest that an autonomous mathematical reconstruction can be used to effectively measure volume loss in retrieved tibial inserts. © 2012 Wiley Periodicals, Inc. J Biomed Mater Res Part B: Appl Biomater 101B: 449–457, 2013.