Random amplified polymorphic DNA (RAPD) analysis using arbitrary 10-mer oligonucleotide primers was employed in order to investigate the genetic fidelity of somatic embryogenesis-derived regenerants of oil palm (Elaeis guineensis Jacq). Clonal palms bearing the "mantled" phenotype were identified in the field and the ability of RAPD markers to distinguish these variants from palms of the normal type was assessed. Of the 387 arbitrary primers used, 259 (67%) were successfully used to amplify oil palm DNA genomic fragments with consistently reproducible banding. Of these 387 primers, 73 (19%) primers enabled the identification of polymorphism between clones. No intraclonal variability and no differences between mother palms and regenerants could be identified using the total number of markers scored (8900). Twenty-four of these 73 primers were chosen for use in a larger experiment aimed at comparing, first, the mother palm genome with that of its clonal offspring and, second, true-to-type and variant regenerants. Thus, the regeneration protocol based on somatic embryogenesis set up for oil palm clonal propagation does not induce any gross genetic changes. The results obtained revealed however, that the RAPD approach is not suitable for the detection of the "mantled" variant phenotype. The use of RAPD markers for the detection of somaclonal variation in oil palm is discussed.