Single cell polymerase chain reaction (PCR) for preimplantation genetic diagnosis (PGD) needs to be highly efficient and accurate. In some single cells from human embryos presumed to be heterozygous for the delta F508 deletion causing cystic fibrosis (CF), we recently observed random amplification failure of one of the two parental alleles following nested PCR. To investigate allele dropout (ADO), we have examined two different lysis protocols and the effect of altering the denaturation temperature in the primary PCR using single lymphocytes heterozygous for delta F508 or for two beta-thalassaemia mutations IVS 1 nt 1 (G/T) and 5 (G/C) using a nested PCR protocol to amplify the 5' region of the beta-globin gene. Amplification rates were high after lysis in either water or lysis buffer and at all denaturation temperatures studied (> or = 92%). With a typical denaturation temperature (93 degrees C), ADO was detected at both loci. When the denaturation temperature was lowered to 90 degrees C, however, ADO increased substantially and conversely by raising the denaturation temperature to 96 degrees C during the first 10 cycles ADO was reduced but not eliminated. ADO was also reduced with cells in lysis buffer. We suggest that ADO may be caused by a combination of inefficient denaturation and degradation of one of the genomic alleles in the first cycles of PCR. For autosomal recessive conditions in which both parents are carrying the same mutation, ADO would not cause serious misdiagnosis. For compound heterozygotes or autosomal dominant conditions, however, extensive testing of the amplification protocol with single heterozygous cells and individual calibration of each thermocycler for the effect of denaturation temperature on ADO is essential before clinical application.