Studies with four different transplantable murine tumors demonstrated that surgical instruments contaminated by contact with a tumor mass could produce tumors in a surgical wound. Eighty-seven per cent of mice with wounds made by invisibly contaminated scissors developed tumors. Irrigation with water did not prevent tumor growth. Before spilled tumor cells can invade and grow into a recurrence in the wound site, they must first attach to underlying extracellular matrix. We have devised a simple in vitro assay to identify inhibitors of tumor-cell attachment to develop therapeutic compounds that can prevent tumor-cell reimplantation. Various test compounds, including proteases (trypsin and Dispase), known modulators of matrix metabolism (proline analogues, cycloheximide, heparin, cortisone, cortexolone, and heparin-steroid combinations), large molecular weight polymers (agarose, dextran, polyethylene oxide), and synthetic fibronectin peptides were tested for their ability to inhibit mouse melanoma (B16-F10) cell attachment to gelatinized dishes. Most of these compounds had little or no effect on tumor-cell adhesion when cells were plated in serum-containing medium. However we identified three compounds that inhibited tumor-cell attachment in a reversible fashion: (1) a specific inhibitor of collagen deposition (L-azetidine-2-carboxylic acid); (2) a bacterial neutral protease (Dispase); and (3) synthetic fibronectin peptides that contained the arginine-glycine-asparate (RGD) sequence that is responsible for cell binding. Dispase and the RGD-containing peptides also inhibited cell implantation and prevented tumor formation in a surgical wound. We propose that inhibitors of attachment might be used either alone or with other biologic modifiers to prohibit implantation of free tumor cells at the time of surgery and thus, to prevent local tumor recurrence.