Purpose To verify that lateral trochlear inclination (LTI) measured by the transepicondylar axis can reliably be used to evaluate trochlear dysplasia (TD) on MRI and can serve as an objective indication of trochleoplasty for patients with lateral patellar dislocation (LPD). Methods Eighty patients with recurrent LPD and eighty healthy subjects were included. TD, posterior condylar angle (PCA), and LTI measured by the posterior condylar reference line (LTIp), surgical transepicondylar axis (LTIs), and anatomical transepicondylar axis (LTIa) were assessed on MRI. The intraclass correlation coefficient (ICC) and Bland-Altman analysis were performed, the correlations and differences amongst the parameters were identified, and a binary logistic regression model was established. Results Each measurement had excellent inter-and intra-observer agreement. The LTIp, LTIs and LTIa were smaller in the study group than in the control group, with mean differences of 9.0°, 7.2° and 7.0°, respectively (P < 0.001). The PCA was larger in patients with LPD than in the control group (P < 0.001). LTIp was associated with PCA in the study group (r = − 0.41, P < 0.001). The pathological values of LTIp, LTIs and LTIa were 11.7°, 15.3° and 17.4°, respectively. LTIs and LTIa were independent risk factors for LPD, with ORs of 7.33 (95% CI [1.06-52.90], P = 0.048) and 10.29 (95% CI [1.38-76.96], P = 0.023), respectively. Conclusion The LTI can be reliably measured by MRI, but LTIp could potentially decrease the recorded value from the actual inclination angle. LTIs and LTIa are more appropriate to serve as trochleoplasty indications for patients with LPD, which could help orthopedists with surgical decision-making. Level of evidence Level III.