Predisposition to coeliac disease (CD) involves HLA genes. We investigated whether any haplotypes modify risk when carried trans to a known high-risk haplotype, DQA1*05-DQB1*02. Earlier attempts to rank levels of risk contributed by the 'other' haplotype were burdened by use of case-control populations; haplotype frequencies were estimated and homozygosity was only presumed. In contrast, exact haplotypes can be determined and allele transmission can be traced in families. A similar study in narcolepsy reported strata of different degrees of predisposition, attributable to the 'other' haplotype. A gene dosage effect similar to that described for DQB1*02 in CD, has also been reported in narcolepsy. We genotyped 439 simplex/multiplex trios for DQA1 and DQB1. We designed a new statistic to test risk modulation by the trans haplotype, even if the affected offspring was homozygous. We tested for significant deviation in transmission of the 'other' haplotype, i.e., modification of DQA1*05-DQB1*02 risk. We also addressed the proposed difference in risk, between DQA1*05-DQB1*02 homozygotes and DQA1*05-DQB1*02/DQA1*0201-DQB1*02 heterozygotes, reported in Southern Europe. We confirmed a DQB1*02 gene dosage effect. However, no haplotypes were found to modify risk when carried trans to DQA1*05-DQB1*02, except DQA1*05-DQB1*02 and DQA1*0201-DQB1*02 which were already known. We did not find credible evidence for a difference in risk conferred by DQA1*05-DQB1*02 and DQA1*0201-DQB1*02, when carried with DQA1*05-DQB1*02. The new test, which directly inspects haplotype transmissions rather than estimated haplotype frequencies, was used to demonstrate that the 'other' haplotype (except DQA1*05-DQB1*02 and DQA1*0201-DQB1*02) does not modify risk conferred by DQA1*05-DQB1*02. The test is applicable to other diseases.