The New York and the Rome diagnostic criteria for ankylosing spondylitis (AS) and the clinical history screening test for AS were evaluated in relatives of AS patients and in population control subjects. The New York criterion of pain in the (dorso) lumbar spine lacks specificity, and the chest expansion criterion is too insensitive. The Rome criterion of low back pain for more than 3 months is very useful. Our study showed the clinical history screening test for AS to be moderately sensitive, but it might be better in clinical practice. As a modification of the New York criteria, substitution of the Rome pain criterion for the New York pain criterion is proposed.