Centre of rotation of angulation (CORA)-based levelling osteotomy (CBLO) is a recent addition to surgical procedures for stabilization of the cranial cruciate ligament-deficient canine stifle joint. Careful identification of the CORA location preoperatively and use of this location intraoperatively are required to ensure accurate correction of the tibial plateau angle. Limited data are available regarding the magnitude and source of potential errors during planning and execution of CBLO. A geometric model enabling isolation of various error sources is described. Landmarks were derived from tibial radiographs (n = 50) by 5 observers and used to define proximal and distal anatomical axes for simulation of CBLO. Observer-specific CORA locations with mean landmark data were used to assess planning errors, and simulated malpositioning of the CORA at 10 mm from the ideal location was used to assess surgical errors. Planning errors result mainly from tibial plateau misidentification, with CORA locations dispersed up to ±10 mm proximodistally from ideal (95% confidence). Malpositioning of the CORA during surgery causes equal and opposite changes in tibial plateau angle (TPA) and anatomical-mechanical axis angles, and varying degrees of translation and limb length changes. The magnitude of these changes is dependent on initial TPA and limb length, with smaller dogs and steeper tibial plateaus resulting in larger errors. Optimal planning and execution are required to achieve the planned outcome of CBLO. The main source of error in our simulation is identification of the tibial plateau. While both pre- and intraoperative errors influenced TPA, based on our geometric model the effect in larger dogs may not be clinically significant. If distalisation of the CORA is required during surgery, compensation of the CORA angle to maintain the target TPA is possible.