The cell culture clonogenic assay appears to have considerable application in predicting individual cancer patient response to chemotherapy. There are several obstacles in the way of its successful application, however, including problems of extrapolating in vitro tests to the in vivo situation, and technical problems related to successful selective in vitro growth of primary human tumour cells. The work in this thesis addresses this latter area, using a human carcinoma line RPMI 2650 as a model system. Different disaggregation methods were compared; after short-term 0 incubation at 37 C, trypsin (0.25%) - EDTA (0.02%) produced the best viable single - cell suspension of RPMI 2650 cells. Following dissociation, RPMI 2650 cells and normal human fibroblasts were found to reaggregate spontaneously to form clusters. Clumps and clusters in cell suspensions contributed to final colony forming efficiencies; the significance of clumps and clusters in the clonogenic assay is discussed. Use of low gelling temperature agaroses improved colony forming efficiencies under some conditions examined. Use of ultrapure water in the preparation of semi-solid media also improved colony formation. Different incubation atmospheres were compared; RPMI 2650 colony formation was improved when the oxygen concentration was reduced from 20% to 3%. Colonies did not, however, grow under anaerobic conditions. Medium without foetal calf serum did not support RPMI 2650 colony formation; growth was also poor in 1% and 2% f.c.s. Inclusion of HITES supplements did not reduce the requirement for serum. Other growth factors tested include TCGF, EGF, insulin and hydrocortisone. Use of feeder cells improved colony forming efficiencies considerably; best results were obtained with mitomycin C-treated feeder cells. Inhibition was observed where high densities of feeder cells were used with high test cell densities. The contribution of cell cycle to final colony formation was examined but did not appear to be a contributory factor to the low cloning efficiencies observed. Normal human fibroblasts did not grow in either agar, agarose or methycelluloses, thus confirming the selective nature of the assay used in this laboratory. Colony formation in agarose was inhibited at a lower ouabain concentration than in monolayer; thioguanine inhibited colony formation in agarose and monolayer at the same toxic level. The significance of these findings is discussed and makes some contribution toward an understanding of the nature of anchorage-independent growth.