Résumé : Diagnosis of genetic disorders in the embryo before implantation, though possible by removal of one or two blastomeres at the eight-cell stage, is still experimental because the procedures of gene analysis of DNA from a single cell are not yet reliable enough for clinical application. We have evaluated the efficiency and accuracy of polymerase-chain-reaction (PCR) amplification of a single-copy gene, wild-type or cystic fibrosis ΔF508 allele, on single sperm cells from a donor known to be a heterozygous carrier of the ΔF508 mutation. DNA from single spermatozoa was decontaminated by restriction-enzyme treatment, then the region around the ΔF508 site was amplified by nested PCR. The distribution of the wild-type and mutant alleles (59 [55%] and 48 [45%, respectively]) in the 107 single spermatozoa did not differ from that expected (50% each). 1 sample did not provide an amplified signal. To check that the two alleles would be amplified with equal efficiency when they were both present within a cell, we did PCR for 51 two-sperm samples. Again the distribution did not deviate from that expected (17 [33%] both wild-type; 21 [41 %] one wild-type, one ΔF508; 13 [26%] both ΔF508 vs 25%; 50%; 25% expected). None of the 74 blanks in these experiments was contaminated. We conclude that our ΔF508 single-cell assay is efficient and accurate and can be used for analysis of blastomere DNA to diagnose cystic fibrosis before embryo implantation. © 1992.