BACKGROUND : Although lateralized head sign (LHS) is considered indicative of structural instability, it lacks reproducibility, and its usefulness in diagnosing adult hip dysplasia is unknown. In this study, we proposed a modified LHS and aimed to determine its reliability and accuracy in diagnosing adult hip dysplasia. METHODS : We reviewed anteroposterior pelvic radiographs of 94 patients (134 hips) with hip dysplasia. As controls, 133 normal hips were evaluated. We defined the modified lateralization distance (LD) as the horizontal distance between the medial aspect of the femoral head to the acetabular fossa. The sensitivity, specificity, and cutoff value of modified LD in diagnosing hip dysplasia were calculated using receiver operating characteristic curve analysis. RESULTS : The sensitivity and specificity of the modified LD (79% and 87%, respectively) were higher than those of the conventional LD (63% and 80%, respectively). Diagnostic accuracy was further improved when patients were divided on the basis of sex. The sensitivity and specificity of the modified LD in men were 89% and 97%, respectively, with a cutoff value of 12 mm, while in women, they were 80% and 96%, respectively, with a cutoff value of 9.7 mm. Therefore, we defined the modified LHS as presentation with a modified LD ≥ 12 mm in men and ≥10 mm in women. Between three observers, the kappa value for intraobserver reliability was ≥0.9, while that for interobserver reliability was >0.8. The mean sensitivity and specificity of the modified LHS in diagnosing hip dysplasia were 85% and 99%, respectively. CONCLUSIONS : Modified LHS-a horizontal distance between the medial aspect of the femoral head and the acetabular fossa ≥12 mm in men and ≥10 mm in women-is a clinically useful marker in detecting lateral femoral head subluxation indicative of hip dysplasia or structural hip instability.