I propose an alternative model of cancer in which metastasis need not all arise out of spread from the "original" tumour. The model assumes that cancer cells arise from stem cells that best grow in the organ of their differentiation. When the internal milieu allows it they also grow at other sites as well, thus complementing the conventional (spreading) metastatic process. Several phenomena in the natural history of cancer, especially breast cancer, that challenge the conventional model, fit well after inclusion of the new model. These are (a) a very modest benefit of screening (b) frequent sparing of lungs from haematogenous metastasis (c) presence of occult cancers in autopsy studies (d) only a modest effect of local treatment (e) relative ineffectiveness of high-dose chemotherapy (f) constant time between surgery and peak of hazard of relapse irrespective of stage of the tumour. All these phenomena are much easier to explain when one rejects the dogma that all metastasis arise only from the primary tumour. This paper is aimed only to suggest an alternative perspective of natural history of solid tumours--to stimulate research on the complex internal milieu that allows cancer cells to develop in new light.