Colorectal cancers carry multiple gene abnormalities at both oncosuppressor and oncogene loci. The role of these defects, singly and in combination with each other, in defining colorectal tumour origin and progression is not known. One means to address this important problem experimentally would be to assess the effects of correcting these gene defects in human tumour cell lines against backgrounds of normality, or defined abnormality at other key loci. Such studies demand the existence of colorectal carcinoma lines in which the major colerectal tumour-associated gene defects are defined. This thesis describes the derivation of 12 such lines and compares them with 6 lines derived elsewhere. Lines derived in this study were propagated firstly as xenografts in immunodeficient mice and briefly as primary cultures. Tumour xenografts were shown to remain stable, faithfully conserving defects or normality, at all sites tested. Stability of the more global index of DNA ploidy was also demonstrated, often over 1-2 years and in one case over 25 monthly passages in vivo. Moreover, of the many lines which grew as primary cultures in vitro (in which genetic manipulation could take place), 2 were subsequently re-implanted in vivo and grew as xenografts with unchanged properties. In contrast to the stability of these xenografts a cell line, intentionally exposed to potential mutagens in vitro, showed a specific rearrangement at the critical locus on 5q21 in association with acquired tumorigenicity. This constitutes the first demonstration of such in vitro effects in direct association with specific APC and/or MCC alterations.