The etiology of a fraction of recurrent spontaneous abortions (RSA) may involve immunological mechanisms. Aberrant profiles of Th1 and Th2 cytokines have been observed which are not present in uncomplicated pregnancies. Studies of classical HLA class I and II antigens in relation to RSA have not been conclusive. Furthermore, these antigens are not expressed in the placenta with the exception of HLA-C. However, HLA-G is expressed on especially invasive cytotrophoblasts and exists in both membrane and soluble forms. HLA-G may be involved in materno-fetal tolerance. Therefore, 61 RSA couples (with three or more spontaneous abortions) and 47 fertile control couples were HLA-G genotyped by direct DNA sequencing and analyzed for specific polymorphisms. No statistically significant differences were observed in the distribution of HLA-G alleles between controls and RSA couples, however, 15% of the RSA women carried the HLA-G*0106 allele compared to 2% of the control women. The 14bp deletion polymorphism in exon 8 was investigated separately. There were a greater number of heterozygotes for the 14bp polymorphism in the group of fertile control women than expected, according to Hardy-Weinberg equilibrium. Furthermore, the HLA-G alleles without the 14bp sequence were prominent in the RSA males in contrast to the RSA women in whom alleles including the 14bp sequence were frequently observed, especially as homozygotes. These results are discussed in relation to two hypotheses concerning HLA-G and RSA. A hypothesis of HLA-G histo-incompatibility between fetus/placenta and the mother was not supported by the data. Another hypothesis concerned certain HLA-G alleles associated with an altered expression profile of HLA-G isoforms or reduced expression of certain HLA-G isoforms.