Objective. To determine which early clinical data differentiate juvenile-onset ankylosing spondylitis (AS) from juvenile rheumatoid arthritis (JRA).Methods. Medical records of 35 patients with juvenile-onset AS and 75 with JRA (excluding type I1 pauciarticular JRA), all of whom had disease onset at age 516 years, disease duration of 52% years at the initial visit to the rheumatology clinic, and followup of 2 10 years, were analyzed retrospectively with regard to features of disease found 6 months, 12 months, and 10 years after onset.Results. At 6 months, various features appeared more frequently in the juvenile-onset AS group than in the JRA group, i.e., pauciarthritis (54.3% versus 30.7%; P = 0.03, odds ratio [OR] = 2.7), enthesopathy (82.9% versus 0%; P < 0.0001, OR = 321.4), tarsal disease (71.4% versus 1.3%; P < 0.0001, OR = 185.0), and lumbar/sacroiliac symptoms (11.4% versus 0%; P = 0.02, OR = 11.9). At 12 months, the features found more frequently among juvenile-onset AS patients than JRA patients were enthesopathy (88.6% versus 4.0%; P < 0.0001, OR = 186.0), tarsal disease (85.7% versus 10.7%; P < 0.0001, OR = 50.3), and knee disease (100.0% versus 82.7%; P = 0.04, OR = 8.0). Involvement of the upper extremities (especially the hands) was found in significantly fewer juvenile-onset AS patients compared with the JRA group. Definite involvement of the spine and sacroiliitis in juvenile-onset AS occurred after a mean & SD of 7.3 f 2.0 years.Conclusion. Regardless of axial disease, enthes-