Prior studies have shown the promise of ultrasound-based approaches to measure knee laxity during clinical laxity assessments. However, a majority of these approaches involve manually measuring the change in the joint gap and/or bone motion before and after applying an external load. These manual measurements make determining changes in kinematics throughout a set of assessments, across many subjects, and/or across many time points impractical. Therefore, the objectives of this study were:​ (1) to develop an ultrasound-based approach to continuously measure bone motion using optical flow, and (2) to compute the errors in measuring knee kinematics during clinical laxity assessments using this approach. Our approach using optical flow allows users to select points along the bone(s) of interest in the first ultrasound frame to automatically define unique regions of interest to track the bone’s motion in successive ultrasound frames. We employed this approach to measure tibiofemoral kinematics using ultrasound B-mode cine loops of five human cadaver knees during anterior, posterior, varus, and valgus clinical laxity assessments. We computed errors in these kinematics compared to those measured using bone-pin-mounted optical motion capture markers. We found the maximum root-mean-square errors of ultrasound-measured kinematics using optical flow to be 1.1 mm and 0.9° for the anterior-posterior and varus-valgus laxity assessments, respectively, when simulating loading rates of clinical laxity assessments. In conclusion, our results demonstrate that our ultrasound-based approach using optical flow is a broadly translatable approach to accurately measure knee kinematics during clinical laxity assessments.