The two-stage model of carcinogenesis, which incorporates clonal growth of intermediate cells, has gained increasing attention in recent years. It was formulated to match tumor incidence data and expanded to encompass observations made in initiation-promotion carcinogenicity experiments. Mouse skin experiments are perceived as supporting this model, with papillomas representing the intermediate cells and carcinomas representing the malignant cells. In this manuscript, the two-stage model is applied to data concerning papilloma and carcinoma formation from an initiation-promotion NMRI mouse skin painting experiment which included stop-promotion. It is shown that the model is not compatible with these data if all papillomas are considered premalignant lesions. The model was modified to allow for a heterogeneous population of papillomas. In this case, unless one assumes that premalignant and terminally benign papillomas are morphologically different in the sense that both types of papillomas at detection limit contain distinct numbers of actively dividing initiated cells, the model predicts larger numbers of papillomas at the end of the experiment than were actually observed. The best explanation is that the model is not in accordance with these data and that the data indicate the need for stages between initiated and malignant cells.