A hypothesis greatly influencing thinking and experimental work in tumor immunology is that many tumor cells arise daily in an organism. However, relatively low numbers of tumours become clinically manifest. This discrepancy has intially led to the hypothesis that the great majority of these potential tumors is killed by immune surveillance mechanisms. After some time, however, serious objections were raised against this function of the immune system. When the interest in immune surveillance faded away, investigators started to study natural resistance against the de novo arising tumor cells. In this paper the data on natural resistance (and immune surveillance) are reviewed. These data lead to the conclusion than an efficient natural resistance that could kill many de novo arising tumor cells is lacking. Taken together the data suggest that no tumors arise when immune surveillance and/or natural resistance are absent. This implies that tumor cells probably do not arise frequently. Acceptance of this conclusion leads to a reappraisal of the role of immune surveillance and natural resistance against tumor cells. This reappraisal will mean a shift from (a) the hypothesized failsafe mechanisms to (b) mechanisms that may or may not kill rarely arising tumor cells, depending for instance on the antigenicity of the tumor cells and their sensitivity to tumoricidal mechanisms.