Three experimental breeding populations, created by pooling four to five natural populations of perennial ryegrass, were used to indicate the effect of sampling and pooling on the retention of genetic variation. Conservation of allelic multiplicity and genotypic frequencies was observed using allelic frequencies measured on six isozyme markers. We observed that by bulking four or five natural populations, represented respectively by 25 or 20 plants, in a large polycross design, the allelic and genotypic frequencies of the original mother-plants were conserved. There was no loss of rare alleles (0.01