Associations between atopic diseases and lymphocyte antigens seem to be possible, but not certain (for a review see McMichael and McDevitt 1977). Several studies (Levine et al. 1972, Marsh et al. 1973, Yoo et al. 1976, Bruce et al. 1976, Scholtz et al. 1976, Seignalet et al. 1976, Marsh and Bias 1977 have shown a certain association between atopy and different HLA antigens or haplotypes, within the same family, but without demonstrating a definite association with one specific antigen or haplotype. Rachelefsky and coworkers (1976) found a strong association between atopic asthma and B lymphocyte group 2, which was not paralleled by an association with the HLA complex. Since hayfever is the classical, representative atopic disease, we present here the HLA typing of 141 patients suffering from Bermuda grass hayfever. These patients comprise two groups: 86 of them were studied during the 1975 season; 65 during the 1976 season; and 10 during both seasons. The diagnosis of hayfever was based on a positive correlation between: (a) family history (several cases of proven atopy in three consecutive generations); (b) clinical picture (seasonal rhinitis, conjunctivitis, and asthma during the Bermuda grass blooming season only); and (c) positive skin tests to whole Bermuda grass extract, obtained from the Hollister-Stier Laboratory. The controls were pooled from data on a recently typed healthy Jewish population (Brautbar et al. 1977, Gazit et al. 1978a, b, c, Zamir et al. 1978) so as to match the stratification of the patients into individuals of Ashkenazi (Russia, Poland, and Roumania) and Sephardic (Marocco and Iraq) origin, in a total of 597 controls for 65 patients.For HLA typing the standard N.I.H. technique was employed (Brand et aL 1970) and the lymphocytes were typed for 9 and 16 antigens at locus A and B, respectively, with 120 alloantisera obtained from the N.I.H. serum bank and from local preparations. The results in both groups (1975 and 1976) show an association with allele B27, at a significance of 1%, and this suggests more than a spurious correlation.The data were subjected to a relative risk analysis. The Haldane (1955) relative risk evaluation uses the formula: