Familial occurrence of coeliac disease (gluten-sensitive enteropathy, GSE) is well known, but the mode of inheritance remains unclear. Pena et al proposed that the genetic basis for GSE was due to disease-predisposing alleles at two loci: DRw3 at the HLA-D locus and a GSE-associated B-cell alloantigen at another, unlinked locus. They concluded that clinical disease required one DRw3 (dominant inheritance) and two B-cell alloantigen alleles (recessive inheritance), but the observed gene frequencies were not consistent with the observed disease prevalence. Here we examine the gene frequencies allowed, assuming a 2-locus model, under the constraints of known disease prevalence limits and the segregation ratio calculated from 42 published families. The gene frequencies found by Pena et al predict the segregation ratio observed in the published pedigrees and the best estimates of disease prevalence of GSE, provided 1) a 2-locus model is assumed and 2) both loci exhibit recessive inheritance. The segregation ratio appears incompatible with a 2-locus dominant-recessive model without assuming reduced penetrance.