The main concern of this thesis is with desmoplasia - a process in which excessive connective tissue is deposited in a neoplasm. This is a common phenomenon in neoplasia but one whose mechanisms are poorly understood. To study the process, I used a human malignant melanoma cell line (UCT-Mel 7) that was established in this laboratory and that, when injected into athymic mice, gave rise to tumours that showed a number of interesting features. Firstly, the tumour induced a marked desmoplastic response as evidenced by a high content of hydroxyproline in tumour lysates, intense staining for reticulin in sections of the tumour and infiltration of the tumour by host mesenchymal cells. Secondly, the desmoplasia was associated in UCT-Mel 7-derived tumours with an unusual phasic pattern of growth that was related to the in vitro passage number of the melanoma cells. On occasions, murine tumours developed at the site of inoculation of human tumour cells. I have identified 2 possible mechanisms by which UCT-Mel 7 cells could have induced the desmoplastic response: either the tumour cells could have exerted their effect indirectly, i.e. via macrophages, or they could have stimulated the host's stromal cells directly. UCT-Mel 7 cells were shown to be chemotactic for mouse macrophages and human foreskin fibroblasts were stimulated, in a dose-dependent manner, to synthesize increased amounts of collagen when co-cultured with mouse peritoneal exudate cells. Stimulation could only be effected by direct cell:cell contact; medium conditioned by macrophages was not effective. The amount of stimulation was not dependent on the state of activation of the peritoneal cells nor on the strain of mouse used. Tumour cells were also found to act directly. Co-culture of UCT-Mel 7 cells and fibroblasts resulted in increased collagen synthesis by the fibroblasts. Increased synthesis of the protein was reflected in an increase in the amount of collagen mRNA. UCT-Mel 7 cell stimulated in a dose-dependent manner with an absolute requirement for intimate cell:cell contact with the fibroblasts. DNA synthesis was not required. Dexamethasone, retinoic acid and the tumour promoter, phorbol myristate acetate, had significant primary effects on fibroblast collagen synthesis but did not modify the response to melanoma cells. Indomethacin, however, had a minimal primary effect upon the fibroblasts but significantly augmented the melanoma cell effect. The nature of the stimulatory cell:cell contact is still uncertain. The gap junction inhibitor, α-glycyrrhetinic acid, did not diminish the melanoma cell effect. Preliminary findings suggested that cell-surface proteoglycans may be important. Removal of the proteoglycans with the inhibitor of proteoglycan assembly, 4-methylumbelliferyl-β-D-xyloside, abrogated the melanoma cell:fibroblast interaction. Recombinant basic fibroblast growth factor did. not seem to be involved in the desmoplastic response. It was of incidental interest to note that this compound inhibited fibroblast collagen synthesis in a manner that was augmented by the concomitant addition of heparin. A surprising finding was the production of a potent inhibitor of collagen synthesis by superinduced cells of the mouse macrophage cell line, P388D₁. This inhibitor has not been fully characterised.