Published data on TP53 mutations can be used to examine the question of whether generalized hypermutability is a necessary condition for tumorigenesis. Although individual mutations do play an etiologic role in tumor formation, the evidence so far does not make it necessary to assume a general mutability. Silent and multiple mutations in the TP53 data set indicate that a special hypermutability process operates on this gene during the generation of tumors. The percentage of silent p53 mutations observed (3%) is at least 20 times greater than would be expected and indicates hypermutability for this gene. The greater proportion of silent mutations among multiple p53 mutations (10%) indicates that the mutations occur nonselectively. The presence of silent mutations implies that not all mutations observed in tumors have an etiologic role. Analysis of the distribution of tumors with two, three, four and more p53 mutations suggests that mutations in some tumors occur in clusters possibly as a result of 'stuttering' in DNA synthesis. It is argued that the most likely alternative explanations of the data, polymorphism and/or a selective role for silent mutations, are not correct. It remains possible that the hypermutability process is restricted to particular genes or to regions of the genome as, for example, in antibody production. There is a surprising paucity of data on human polymorphism and nucleotide diversity which makes the analysis difficult.