Objective. Using humoral immune responses, Klebsiella pneumoniae has been implicated as a candidate microbial trigger in ankylosing spondylitis (AS) by several investigators but refuted by others. The objective of this case-control study was to compare the cellular (T-cell proliferation) and humoral (IgG and IgA, by ELISA) immune responses of affected individuals in multiplex AS families with those of unaffected family members and normal healthy controls in order to find out whether affected individuals exhibit a predominant immune response to K. pneumoniae. Methods. Twenty-five families with two or more individuals affected with AS and 34 normal healthy controls matched with the affected family members for age, sex and ethnicity were enrolled in the study. All affected (n = 57) and unaffected (n = 37) family members had a detailed clinical evaluation. Peripheral blood was drawn to determine T-lymphocyte proliferation and the IgG and IgA (by ELISA analysis) immune responses to K. pneumoniae, Salmonella typhimurium, Yersinia enterocolitica and Chlamydia trachomatis. Immune responses to each of the four candidate organisms were compared in affected and unaffected individuals. Each individual was classified by the predominant antigenic immune response that they showed when comparison was made among the same concentrations of the four candidate microbial antigens. This stratification was then used (i) to compare immune responses in affected and unaffected family members and (ii) to compare clinical characteristics of affected family members. Results. There was no difference in mean stimulation indices or antibody responses between affected and unaffected family members for each of the candidate organisms. In terms of predominant cellular immune responses to these organisms, there was no difference between affected and unaffected family members with respect to K. pneumoniae, C. trachomatis or Y. enterocolitica. However, a higher percentage of affected family members (25.9%) exhibited a predominant response to S. typhimurium compared with unaffected family members (5.9%, P < 0.02). In assessing antibody titres, K. pneumoniae was the predominant amongst these four organisms, but there was no difference between affected family members, unaffected family members and normal healthy controls. There was no relationship between immune responses and clinical characteristics. Conclusion. Our analysis of affected and unaffected family members in familial AS demonstrated no significant differences with respect to cellular or humoral immune responses to K. pneumoniae and three control microbes. In addition, K. pneumoniae did not exhibit a predominant immune response in affected individuals. Thus we find no supportive evidence to implicate a causal role for K. pneumoniae in familial AS.