PurposeAbnormal patellar height has been identified as a source of aberrant mechanical functioning within the patellofemoral joint. The purpose of this study is to examine the statistical agreement among three commonly used classification methods: Blackburne‐Peel (BPI), Caton‐Deschamps (CDI) and Insall‐Salvati (ISR), by evaluating (1) the rates of patella alta identification and (2) the ability for one index to predict another.MethodsOne hundred lateral knee radiographs were evaluated using BPI, CDI and ISR to classify each knee as patella normal, patella alta or patella baja. Linear regression analysis was performed to evaluate the relationship between each index. Conversion equations were then derived using the reported linear regression best‐fit line, comparing each pair of indices.ResultsPatella alta was identified in 15 knees using BPI, 15 using CDI and 25 using ISR. A total of seven knees were classified as patella alta by all BPI, CDI and ISR. Statistical analysis revealed significant correlation (p ≤ 0.001) among BPI and CDI (R2 = 0.706), BPI and ISR (R2 = 0.328) and CDI and ISR (R2 = 0.288). Wilcoxon Signed‐Rank test between the three indices revealed no significant difference between the means of converted and original indices.ConclusionDespite their significant correlations and adequate reproducibility, variability between common patellar height indices render predictions and conversions between BPI, CDI and ISR inequivalent. Users of these indices must be aware of their incongruent properties when considering application to patients in the clinical setting. Furthermore, it remains unclear which patellar height measurement technique is the correct index to use in a given knee. This study highlights the need for further investigation to create a reliable and standardised method for identifying patella height.Level of EvidenceLevel IV.