PURPOSE:To develop methods for evaluating 3D patellofemoral and tibiofemoral alignment in vertical open-bore magnetic resonance (MR) scanners, with participants upright and fully weight-bearing; and to evaluate the repeatability of these methods in individuals with patellofemoral osteoarthritis (OA) and in asymptomatic knees. MATERIALS AND METHODS:Our methods extend previously validated, reliable methods for evaluating alignment into an upright MR environment. In 10 participants with early patellofemoral OA and 10 with asymptomatic knees, we acquired sagittal T1 -weighted turbo spin echo images in a 3T scanner to create accurate participant-specific 3D anatomical surface models. In a vertical open-bore 0.5T MR scanner, we obtained lower-resolution sagittal gradient echo images to capture bony position and orientation data. Participants were scanned in a position of squatting with the knees flexed 30°, three separate times to evaluate repeatability. Bone segmentation was performed manually, surface models were registered to data from the 0.5T scanner, and 3D patellofemoral and tibiofemoral alignment was calculated in all six degrees of freedom (three rotations and three translations). RESULTS:Intraclass correlation coefficients (ICCs) were ≥0.94, with the exception of patellar spin (0.79). Standard errors of measure (SEM) were <2° rotation and <0.9 mm translation. Repeatability remained adequate when stratified by group, with the exception of patellar spin (ICC 0.57 for asymptomatic knees vs. 0.91 for OA knees). CONCLUSION:We demonstrate methods for evaluating 3D alignment in upright fully weight-bearing participant positions in a vertical open-bore MR scanner. With the exception of patellar spin, repeatability was good to excellent. LEVEL OF EVIDENCE:3 Technical Efficacy: Stage 1 J. Magn. Reson. Imaging 2018;47:841-847.