Background: Research on the cause of lateral patellar dislocation (LPD) has focused on trochlear morphologic parameters, joint alignment, and patellofemoral soft tissue forces. A paucity of information is available regarding how patellar morphologic parameters influence the risk for LPD. Purpose/Hypothesis: The purpose was to assess whether patellar morphology is a risk factor for recurrent LPD. It was hypothesized that (1) patients with recurrent LPD would have decreased patellar width and volume and (2) patellar morphologic parameters would accurately discriminate patients with recurrent LPD from controls. Study Design: Cohort study (diagnosis); Level of evidence, 3. Methods: A total of 21 adults with recurrent LPD (age, 29.7 ± 11.1 years; height, 170.8 ± 9.9 cm; weight, 76.1 ± 17.5 kg; 57% female) were compared with 21 sex- and height-matched controls (age, 27.2 ± 6.7 years; height, 172.0 ± 10.6 cm; weight, 71.1 ± 12.8 kg; 57% female). Three-dimensional axial fat-saturated magnetic resonance imaging scans were used to measure patellar medial, lateral, and total width; patellar volume; patellar medial and lateral facet length; the Wiberg index; and previously validated knee joint alignment and femoral shape measurements (eg, tibial tuberosity to trochlear groove distance, trochlear dysplasia). Results: The LPD group demonstrated reduced medial patellar width (Δ = −3.6 mm; P < .001) and medial facet length (Δ = −3.7 mm; P < .001) but no change in lateral width or facet length. This resulted in decreased total patellar width (Δ = −3.2 mm; P = .009), decreased patellar volume (Δ = −0.3 cm3; P = .025), and an increased Wiberg index (Δ = 0.05; P < .001). No significant differences were found for all other patellar shape measures between cohorts. Medial patellar width was the strongest single discriminator (83.3% accuracy) for recurrent LPD. Combining medial patellar width, patellofemoral tilt, and trochlear groove length increased the discrimination to 92.9%. Conclusion: The medial patellar width was significantly smaller in patients with recurrent LPD and was the single most accurate discriminator for recurrent LPD, even compared with traditional trochlear shape and joint alignment measures (eg, trochlear dysplasia, patella alta). Therefore, medial patellar morphology should be assessed in patients with LPD as a risk factor for recurrence and a potential means to improve treatment.