Ehlers-Danlos syndrome (EDS) types I and II, which comprise the classical variety, are well characterized from the clinical perspective, but it has been difficult to identify the molecular basis of the disorder in the majority of affected individuals. Several explanations for this failure to detect mutations have been proposed, including genetic heterogeneity, failure of allele expression, and technical difficulties. Genetic heterogeneity has been confirmed as an explanation for such failure, since causative mutations have been identified in the COL5A1, COL5A2, and tenascin X genes and since they have been inferred in the COL1A2 gene. Nonetheless, in the majority of families with autosomal dominant inheritance of EDS, there appears to be linkage to loci that contain the COL5A1 or COL5A2 genes. To determine whether allele-product instability could explain failure to identify some mutations, we analyzed polymorphic variants in the COL5A1 gene in 16 individuals, and we examined mRNA for the expression of both alleles and for alterations in splicing. We found a splice-site mutation in a single individual, and we determined that, in six individuals, the mRNA from one COL5A1 allele either was not expressed or was very unstable. We identified small insertions or deletions in five of these cell strains, but we could not identify the mutation in the sixth individual. Thus, although as many as one-half of the mutations that give rise to EDS types I and II are likely to lie in the COL5A1 gene, a significant portion of them result in very low levels of mRNA from the mutant allele, as a consequence of nonsense-mediated mRNA decay.