Background A lateralized tibial tubercle is one potential cause of patellar instability. The tibial tubercle-trochlear groove (TT-TG) distance using CT is a reliable measure and considered the gold standard. Using MRI for this purpose has increased, although the reliability of doing so is not well studied. Questions/purposes We sought to (1) determine variability in the insertion of the patellar tendon relative to the tibial tubercle and whether this affects the measurement on MRI of the traditional TT-TG distance versus the functional patellar tendon-trochlear groove (PT-TG) distance, (2) determine the reliability of measuring the osseous TT-TG distance, (3) determine the reliability of measuring the soft tissue PT-TG distance, and (4) compare the reliabilities of using osseous (TT-TG) versus soft tissue (PT-TG) landmarks. Methods Four observers measured the TT-TG and the PT-TG distances of 50 MR images of knees obtained for any reason. Each observer repeated these measurements 30 days later. The interobserver and intraobserver reliabilities, measurements per observer that varied from the group mean by greater than 2 mm, and the limit of agreement were calculated. Results The TT-TG and PT-TG differed by as little as 0.11 mm and by as much as 4.18 mm with an average difference of 1.37 mm. The interobserver and intraobserver reliabilities were greater than 90% for the PT-TG and TT-TG distances. The PT-TG distance was less variable in that this measurement showed interobserver and intraobserver reliabilities of 0.977 and 0.972 respectively, versus 0.913 and 0.961 for the TT-TG measurement. Additionally, the PT-TG measurements resulted in a lower average difference to the mean for each observer, less number of knees per observer where the difference to the mean was greater than 2 mm, and improved limit of agreement. Conclusions The TT-TG and the PT-TG distances were not identical and differed by as much as 4.18 mm; as such they are not interchangeable when measuring this distance.