The modified New York criteria for the diagnosis of ankylosing spondylitis were evaluated and compared to the older criteria in 151 patients, referred to hospital because of low back pain and who had a positive clinical history screening test for ankylosing spondylitis and in 31 controls with non-inflammatory back pain. Radiological examination of the sacro-iliac joints showed sacro-iliitis in 124 (82%) from the 151 with inflammatory back pain. In 110 (72%) of those patients a diagnosis of definite ankylosing spondylitis according to the classical New York criteria could be made and they had a prevalence of HLA-B27 of 84%. Application of the modified New York scheme increased the number of patients meeting the criteria for definite ankylosing spondylitis to all 124 patients with sacro-iliitis, and 82% of this group carried HLA-B27. The classical New York criteria of 'limitation of the lumbar spine in three directions' and of 'limitation of chest expansion' appeared to reflect disease duration rather than help in the initial diagnosis.