During a systematic search for germ-line APC mutations causative of familial adenomatous polyposis, we discovered what appeared to be an insertion mutation while simply checking exon 14PCR products by agarose gel electrophoresis (AGE). On AGE, exon 14PCR product from the known affected member of this family gave two bands: one of normal length, the other retarded on the gel equivalent to an increase in length of some 20-25 bp. Direct sequencing of DNA purified from the two bands gave identical results, and was consistent with amplification from the same two alleles: one wild-type, and the other having an 1893del4 mutation. This suggested that the normal length band on AGE consisted of DNA homoduplexes (normal:normal and mutant:mutant) and the retarded band consisted of DNA heteroduplexes (normal:mutant and mutant:normal). This hypothesis was tested by subjecting purified material from each of the two bands alone to a single cycle of heat denaturation and annealing, which showed that either band was equally capable of regenerating both bands. Because the anomalous migration of the heteroduplexes is observed in the presence of ethidium bromide, it implies that they have a cruciform of cruciform-like structure. This case illustrates the necessity to be aware of anomalous DNA migration and always sequence all putative mutations.