Diagnosis, treatment and follow-up of children with juvenile idiopathic arthritis are based upon the articular examination. The objective of the current study was to assess the interobserver agreement in grading the joint with arthritis and limited range of motion, and to detect the sources of disagreement. Twenty children with juvenile idiopathic arthritis were included. Assessment parameters were a) Physician Global Assessment, b) number of joints with active arthritis, and c) number of joints with limited range of motion. A total number of 1,320 joints were examined by both observers, blinded to each other. Each observer marked the involved joints as active arthritis, as limited range of motion, or as both. Interobserver agreement was determined by using Kappa statistics and Wilcoxon signed rank test. The statistical analysis was performed by SPSS 10.0. There was no statistical difference in the comparison of Physician Global Assessment of the two observers. There was substantial agreement in both arthritis and limited range of motion scores. Chance corrected agreement for the interobserver reliability was also performed for frequently involved joint such as neck, wrist, hip and knee. A substantial agreement was found. The close agreement between the observers could be attributed to standardization of examination techniques and to using objective variables.