The insulin-like growth factor II (IGF2) gene is imprinted with the paternal allele expressed and the maternal one silent. Loss of imprinting (LOI) of IGF2 has been suggested to play a role in the development of tumours, but the reported incidence of IGF2 LOI in tumours shows considerable variation, which may stem from different methodologies employed. In particular, partial digestion of reverse transcriptase-polymerase chain reaction (RT-PCR) products by restriction enzymes can lead to inaccurate measurements. To overcome the problem of partial enzymatic digestion, a novel method termed allele specific-polymerase chain reaction (AS-PCR) has recently been reported, which provides a significant advance over enzymatic digestion. A second problem with measurements of biallelic IGF2 transcription is that the co-amplification of contaminating genomic DNA during the RT-PCR step can lead to an overestimation of the frequency of biallelic IGF2 expression. To investigate the extent of this problem, total RNA from breast and colorectal cancer was analysed using two methods. The first method involved a first-round PCR using cDNA generated with primers spanning exons 8 and 9 (exon connection), followed by a second round of AS-PCR using primers from within exon 9. The second method used only AS-PCR with primers from within exon 9. The result was that the exon-connection approach was more accurate, thereby highlighting a significant problem in imprinting analyses where genomic DNA contamination cannot be completely ruled out.