In the aim of better understanding molecular phenomena underlying the "mantled" somaclonal variation in oil palm, our research work has focused on the role of genomic DNA methylation. Two complementary methods, aimed at evaluating the methylation rate at the genome-wide scale, have previously demonstrated the occurrence of a highly significant DNA hypomethylation in leaves and calli from abnormal regenerants, compared to their normal counterparts. Investigations are now aimed at targeting the individual sequences which epigenetic misregulation could account for the onset or for the maintenance of the "mantled"' phenotype. We are currently trying, through a wide range of techniques, to focus our methylation studies on the only genes that are relevant with regard to the "mantled" phenotype. In this aim, we have undertaken methylation-sensitive RFLP and AFLP studies, involving the isoschizomeric enzymes Mspl and HpaII. RFLPs were envisaged primarily in order to screen a pool of oil palm cDNA, while the aim of MSAP (methylation-sensitive amplified polymorphism) investigations was to generate a large number of relevant markers, exhibiting a differential methylation pattern depending on the normal/mantled phenotype but independent of the genetic origin of clones. Hypomethylating drugs were used in embryogenic cell suspensions in order to evaluate the impact of DNA demethylation on the genotype of regenerants and to simultaneously assess the set of methylation-sensitive markers previously developed. By studying oil palm DNA-methyltransferases, it will be possible to determine whether DNA methylation is the only epigenetic regulator affected, and which regions of the genome are specifically modified in variants.