The purpose of this study was to assess the accuracy of femoral tunnel placement in a medial patellofemoral ligament reconstruction (MPFL-R) cohort. The secondary purpose was to establish the evidence of a learning curve to achieve acceptable femoral tunnel placement during MPFL-R. Two surgeons, using lateral radiographs, assessed 73 subjects post-MPFL-R. Femoral tunnel accuracy and direction of tunnel error were measured in relation to Schöttle's point (A-T distance). Interrater reliability (intraclass correlation coefficient 2,k) of these measures was calculated. Learning curve of accurate femoral tunnel placement was examined by dividing the patient cohort into quartiles. A one-way analysis of variance was used to assess the quartiles for accuracy of femoral tunnel position and surgical time. In relation to Schöttle's point, 66/73 (90.4%) femoral tunnels were categorized as being in a “good” or “excellent” position and 7/73 (9.6%) were categorized as being in a “poor” position. Evidence of an MPFL-R learning curve was established via a statistically significant difference in the mean A to T distance for the four quartiles (F [3, 69] = 6.41, p = 0.001). There was also a statistically significant difference in the surgical time for the four quartiles (F [3, 69] = 8.71, p = 0.001). In this series, accurate femoral tunnels were placed more than 90% of the time during MPFL-R. A clear learning curve for accurate femoral tunnel placement was demonstrated both with respect to distance of the tunnel from Schöttle point and with regard to surgical time. Level of evidence was IV.