It is generally agreed that most malignancies, particularly those that arise "spontaneously", are caused by randomly occurring mutations at specific sites of the genome. Hence oncogenesis of these spontaneous tumors can be described by stochastic mathematical models. In this paper we offer a mathematical approach to oncogenesis. A stochastic model was developed to calculate the number of mutations required for malignant transformation. This model demonstrates that the two hit model, as originally proposed by Knudson for retinoblastoma in children, is not tenable for tumors in adults. Our results show that malignant transformation is more likely to be due to a specific set of four mutations. This stochastic model is compatible with most current views on oncogenesis and most phenomena in oncology.