Summary. Measurement of joint health is critically important when assessing children with haemophilia. Few measures exist; they lack sensitivity to small changes, don't account for normal development and were never formally validated. To address these concerns, the Hemophilia Joint Health Score (HJHS) was developed by modifying existing scores. Objective: To test the inter-observer and test–retest reliability of the HJHS. Methods: Using a fully factorial design, four physiotherapists (from Canada, the United States and Sweden) examined eight boys with severe haemophilia A on two consecutive days using the HJHS. The boys ranged in age from 4–12 years and presented with variable joint damage. Six index joints (elbows, knees and ankles) were assessed on 11 impairment items including swelling, flexion and extension loss and gait. Concordance was measured by the intra-class correlation co-efficient. Results: Reliability of the HJHS was excellent with an inter-observer co-efficient of 0.83 and a test–retest of 0.89. Conclusion: This study is the first in a series to assess the psychometric properties of the HJHS, a promising new measure of joint health in boys with haemophilia.