The aim of this study was to evaluate the clinical features, evolution and reliability of spondyloarthropathy criteria in a subset of patients with subclinical sacroiliitis and inflammatory bowel disease (IBD). All patients with IBD (n 62) attending a gastroenterology clinic from a referral centre were included to assess the prevalence of articular involvement. Patients were evaluated according to a specific protocol designed for the study, which included epidemiological and clinical variables, physical examination and radiological assessment. Only those with subclinical sacroiliitis were followed prospectively for 4 years. This group was visited every 6 months with the same initial protocol. Sacroiliac joints were studied using frontal and oblique X-ray views and graded according to New York criteria. HLA B27 typing was performed by serological methods in all patients and in 80 healthy controls. The reliability of Amor and ESSG criteria for spondyloarthropathy was evaluated. Fifteen patients (24%) presented with isolated subclinical sacroiliitis. In this group a higher frequency of peripheral arthritis and erythema nodosum was observed (p = NS compared to those without sacroiliitis). Most cases (60%) were grade II unilateral sacroiliitis. Three patients were HLA B27+ (p>0.05 compared to healthy controls). The resultant sensitivity of Amor's and ESSG criteria ranged from 40% to 46%. An unexpectedly high freuqency (9.5%) of psoriasis was observed in the whole group. There is a high prevalence of isolated subclinical sacroiliitis in IBD. This may represent a forme fruste of enteropathic ankylosing spondylitis, a stunted form of axial involvement because of therapy, or a third category of rheumatic disease associated with IBD. It may also represent a common characteristic of spondyloarthropathies, rather than a specific finding of IBD. The recently developed spondyloarthropathy criteria are not particularly helpful for the diagnosis of this milder form of spondyloarthropathy.