Preimplantation genetic diagnosis (PGD) can be offered as an alternative to prenatal diagnosis (PND) to couples at risk of having a child with a genetic disease. The affected embryos are detected before implantation by fluorescent in situ hybridisation (FISH) for sexing (X‐linked diseases) and chromosomal disorders (numerical and structural) or by polymerase chain reaction (PCR) for monogenic disorders (including some X‐linked diseases). The accuracy and reliability of the diagnosis is increased by analysing two blastomeres of the embryo. However, the removal of two blastomeres might have an effect on the implantation capacity of the embryo. We have evaluated the implantation of embryos after the removal of one, two or three cells in 188 PGD cycles where a transfer was done. The patients were divided into five groups: a first group which received only embryos from which one cell had been removed, a second group which received only embryos from which two cells had been removed, a third group which received a mixture of embryos from which one and two cells had been taken, a fourth group where two and three cells had been removed, and a fifth group where three cells had been removed. The overall ongoing pregnancy rate per transfer was 26.1%, the overall implantation rate per transfer was 15.2% and the overall birth rate was 14.2%. Although pregnancy rates between the groups cannot be compared because the second group (two cells removed) contains more rapidly developing and therefore ‘better quality’ embryos, an ongoing pregnancy rate of 29.1% and an implantation rate of 18.6% per transferred embryo in this group is acceptable, and we therefore advise analysing two cells from a ≥7‐cell stage embryo in order to render the diagnosis more accurate and reliable. Copyright © 2000 John Wiley & Sons, Ltd.