Screening for oncogene mutations as a marker for malignancy can be a powerful tool for the early diagnosis of cancer. The enrichment polymerase chain reaction (PCR) is a sensitive method for the detection of low-frequency mutations in small samples. However, false-positive results, caused by methodological errors, may have severe clinical implications. When applied to the detection of Ki-ras mutations in pancreatic secretions, the assay sensitivity is limited to approximately 1:1400. Our investigation of Ki-ras mutations in blood samples from patients with pancreatic carcinoma revealed PCR bands presumably derived from mutant Ki-ras in samples from healthy volunteers, while all blood samples of the patients with pancreatic carcinomas showed a wild-type band pattern. Mathematical modeling of the PCR reaction reveals that the rate of false positive PCR results depends on the initial amount of DNA, the Taq polymerase error rate, the number of PCR reaction cycles, reaction efficiency and the restriction endonuclease chosen. The overall error rate of false positive results of the enrichment PCR can be reduced to the square of the rate of a single-step analysis if repeated amplifications of the same DNA specimen show an identical result.