This study analyzes a set of dose-response curves relative to promoter carcinogens administered to experimental animals with or without initiator pretreatment. In the case of initiator plus promoter treatment, as well as for low experimental doses, the experimental data and their mathematical fitting indicate a downward shape of the dose-response curves as a general feature. In this study, the following hypothesis, according to the two-stage carcinogenic action of initiator plus promoter treatment, is presented to interpret this downward shape: (i) an initiating action, involving a subset of available "targets" (the number of "initiated" targets is assumed to depend on initiating treatment level), and (ii) a subsequent dose-dependent promoting process, mainly involving the initiated targets. This process reaches a saturation level when all the initiated targets have been promoted. This kind of process may be quantitatively described by a saturated exponential model: P(d) = K(1 - exp(-K0 + K1d)] which fits the experimental data fairly well. No indication supporting the hypothesis of a threshold for the chemicals examined resulted from the analysis. In the specific case of sodium saccharin, experimental dose-response relationships were available with or without initiator pretreatment (BBN and MNU). While in the first case the shape of the dose-response curve is downward, in the second case, it is upward and significantly different.