The quantitative use of the polymerase chain reaction (PCR) is often compromised by the variability of the amplification. The most useful system for quantitation by PCR involves the use of controls which are almost identical to the target and which can be amplified using the same primers as the sequences of interest. In this paper, we use a model system consisting of differently sized targets amplifiable with varying primers to demonstrate the effects of the plateau phase of PCR on quantitation by PCR. This model confirms two commonly observed results: (i) when varying amounts of a single target are amplified, a constant maximum level of product is obtained and (ii) coamplification of different concentrations of different targets results in retention of the initial proportions. The inherent contradiction in these results is examined by replacement of the key elements of the reaction including enzyme, dNTPs or primers, none of which have an effect on the plateau. Pyrophosphate is found to exert no inhibitory effect on the reaction, nor does the exonuclease action of the enzyme cause the plateau. Levels of amplification attained during amplification are both theoretically and empirically defined as being insufficient to lead to the plateau due to competition between self-annealing of product DNAs and primer binding. We conclude that, pending further biochemical enquiry into the enzyme(s) used in the PCR, none of reasons conventionally proposed for the plateau phase of the PCR are sufficient to explain the phenomenon. This being so, we define the plateau as being a feature of the reaction as a whole and, since the onset of this phase is simultaneous for all amplicons, quantitation using the internal control system need not require exponential amplification. This therefore greatly simplifies the quantitative application of PCR.