The extent of chromosomal mosaicism in human preimplantation embryos was examined using an improved procedure for the preparation and spreading of interphase nuclei for use in fluorescence in situ hybridisation, allowing the analysis of every nucleus within an embryo. One cell showed no hybridisation signals in only three of the 38 embryos that were included in this study, i.e. the hybridisation efficiency per successfully spread nucleus was 99% (197/200). Double-target in situ hybridisation analyses with X- and Y-chromosome-specific probes was performed to analyse nine embryos resulting from normal fertilisation, 22 polypronucleate embryos and seven cleavage-stage embryos where no (apronucleate) or only one pronucleus (monopronucleate) was observed. We also analysed autosomes 1 and 7 by double-target in situ hybridisation in the nuclei of two apronucleate, one monopronucleate and four polypronucleate embryos. All nine embryos that resulted from normal fertilisation were uniformly XY or XX. None of the apronucleate or monopronucleate embryos was haploid: three were diploid, one was triploid and three were mosaic. Fertilisation was detected by the presence of a Y-specific signal in four of these embryos. Of the polypronucleate embryos, two were diploid, two were triploid and 18 were mosaic for the sex chromosomes and/or autosomes 1 and 7. These results demonstrate that fertilisation sometimes occurs in monopronucleate embryos and that chromosomal mosaicism can be detected with high efficiency in apronucleate, monopronucleate and polypronucleate human embryos using fluorescence in situ hybridisation.