Radiologic hip surveillance is recommended for children with cerebral palsy (CP) at risk of hip displacement. Young children with abnormal proximal femoral geometry (Hilgenreiner epiphyseal angle, HEA) may be more likely to develop hip displacement, less likely to respond to nonsurgical intervention, and may benefit from earlier surgical referral. The reliability of radiographic measures of migration percentage (MP) in the immature pelvis of young children has been reported in smaller retrospective studies; HEA has not been examined in this population. This prospective study describes the reliability of MP and HEA in very young children with CP. Participants were entered from tertiary referral center CP clinics into a prospectively recruited population-based cohort for hip surveillance with pelvic radiography using standardized patient position, at 18, 24, 30, 36, and 48 months. All Gross Motor Function Classification System (GMFCS) levels were included. Two independent raters assessed radiographs for HEA and MP. The intraclass correlation coefficient (ICC) was computed as a measure of interrater and intrarater reliability. The correlation coefficient between HEA and femoral position was computed. Ninety-eight children less than 25 months (spasticity=83, 85%; GMFCS IV-V=38, 39%), and 114 children 25 to 48 months (spasticity=96, 85%; GMFCS IV-V=37, 32%) were included from 133 unique participants (spasticity=111, 84%; GMFCS IV-V=42, 32%). Of these 79 children were studied in both age groups. Overall interrater and intrarater reliability of MP was high [ICC=0.93; 95% confidence interval (CI), 0.91-0.95]; SEM was 3.9% (single) and 5.5% (sequential). Perfect concordance for classification of marked hip displacement (MP>30%) occurred in 217 cases (95.2%); nonweighted κ=0.80; 95% CI, 0.68-0.91. For HEA, overall reliability was high (ICC=0.89; 95% CI, 0.85-0.93); SEM=4.8% (single) and 6.7% (sequential). Correlation between changes in HEA and femoral abduction was poor (coefficient=-0.27, P=0.244). MP and HEA can be reliably applied to very young children with CP, with high reliability for both measures. Measured HEA values appear to be independent of patient position, and may reflect genuine changes in proximal femoral geometry. A longitudinal study should be performed to determine the relationship between HEA and later hip outcomes. Level I/II--testing and development of diagnostic tests.