We developed a general method of quantifying relative copy numbers of specific DNA sequences based on the theoretical accumulation of polymerase chain reaction (PCR) products when two DNA sequences are amplified together (co-amplified). Our experiments illustrate the development and theory of the technique. The precision of our estimates is demonstrated by statistical confidence intervals. Tests for effects introduced by experimental factors were performed. The precision of the technique was established by examining the relative gene dosage of the X-linked dystrophin gene in human genomic DNAs from a male, a normal female, a 47,XXX female, and a 48,XXXX cell line. The sensitivity was sufficient to distinguish three copies of the gene from four copies; equivalent to detecting loss of heterozygosity in half the cells of a tumour. Confidence intervals allowed us to reject the hypothesis that there was no difference between DNA samples. Four sample pairs would be required to demonstrate relative gene dosage ratios of 2.0 to 1.0; eight sample pairs would be required to demonstrate a relative gene dosage ratio of 1.3 to 1.0. This method should be useful in detecting gene amplification and deletion in a variety of situations.