1 report a striking association between a dinucleotide polymorphism in the interferon gamma gene (IFNG) and rheumatoid arthritis in Canadians. In a commentary, W E R Ollier (Sept 2, p 783) 2 raised the question of whether their allele typing could have slipped by one position, which would lead to the highly significant differences between cases and controls. We did an extensive analysis of this polymorphism in 1730 white European individuals. 3 Since the primers used for amplification of the IFNG poly- morphism in our and Khani-Hanjani and colleagues' studies were identical, the numbers of CA repeat units per allele have been determined unambiguously. Direct comparison of IFNG allele carriage rates (ie, the proportion of individuals carrying a certain allele) between the two studies is possible and scientifically correct. We show in the table the IFNG allele carriage rates in German, Swedish, and Sardinian healthy individuals in our study, compared with those for the Canadian white healthy controls of Khani-Hanjani and colleagues. In Europeans, the 122 bp IFNG allele, which encompasses 11 CA repeat units, is rare. The carriage rate of the 122 bp allele was not higher than 1% in any population in our study. According to Khani-Hanjani and colleagues, the carriage rate of this allele is 80% in the white Canadian Canadian (n=65)* 52 (80%) 44 (68%) 8 (12%) 8 (12%) 0 0 Canadian (n=65)† 0 52 (80%) 44 (68%) 8 (12%) 8 (12%) 0 Sardinian (n=413) 0 326 (79%) 257 (62%) 71 (17%) 26 (6%) 0 German (n=367) 3 248 (68%) 251 (68%) 46 (12·5%) 21 (6%) 14 (14%) Swedish (n=54) 0 32 (59%) 35 (65%) 9 (17%) 6 (11%) 0 *Original assignment of alleles from Khani-Hanjani and colleagues. †Reassignment of alleles from Khani-Hanjani and colleagues.