Ankylosing spondylitis is diagnosed once or twice in each 1000 males and one tenth as frequently in females, but the true prevalence is unknown. Indentification of genetic marker, HL-A W27, for susceptible persons has provided a tool facilitating epidemiologic studies and allowing identification of "control" populations without the marker. Evaluation by postal questionnaires, and pelvic radiography of 78 HL-A 27W-positive blood donors selected from a group of apparently healthy subjects revealed 14 who satisfied the criteria for definite ankylosing spondylitis. The prevalence was similar in both sexes. One hundred and twenty-six W27-negative controls matched for race, sex, and age failed to yield a single case. For a person of either sex with HL-A W27, there appears to be about a 20 per cent chance that ankylosing spondylitis will develop, suggesting a prevalence of 10 to 15 per thousand. Hitherto accepted figures may underestimate the frequency by a factor of 10 to 20.