Abstract: Analysis of polyethylene (PE) wear in knee implants is crucial for understanding the factors leading to revision in total knee arthroplasty. Importantly, current experimental and computational methods for predicting insert wear can only be validated against true in vivo measurements from retrievals. This study quantitatively investigated in vivo PE wear rates in fixed-bearing (FB) (n = 21) and rotating-platform (n = 53) implant retrievals. 3D surface geometry of the retrievals was measured using a structured light scanner. Then, a reference surface that included the deformation, but not the wear that the retrievals had experienced in vivo, was constructed using a fully automatic surface reconstruction algorithm. Finally, wear volume was calculated from the deviation between the worn and reconstructed surfaces. The measurement and analysis techniques were validated and the algorithm was found to produce errors of only 0.2% relative to the component volumes. In addition to quantifying cohort-level wear rates, the effect of mechanical axis limb alignment on mediolateral wear distribution was examined for a subset of the retrievals (n = 14 + 26). Our results show that FB implants produce significantly (p = 0.04) higher topside wear rates (24.6 +/- 10.1 mm3/year) than rotating-platform implants (15.3 +/- 8.0 mm3/year). This effect was larger than that of limb alignment, which had a smaller and nonsignificant influence on overall wear rates (+4.5 +/- 11.6 mm3/year, p = 0.43). However, increased varus alignment was associated significantly with greater medial compartment wear in both the FB and rotating-platform designs (+1.7 +/- 1.3%/degrees and +1.8 +/- 1.6%/degrees). Our findings emphasize the importance of implant design and limb alignment on wear outcomes, providing reference data for improving implant performance and longevity.