The biological activity of scaffolds used in tissue engineering applications hypothetically depends on the density of available ligands, scaffold sites at which specific cell binding occurs. Ligand density is characterized by the composition of the scaffold, which defines the surface density of ligands, and by the specific surface area of the scaffold, which defines the total surface of the structure exposed to the cells. It has been previously shown that collagen-glycosaminoglycan (CG) scaffolds used for studies of skin regeneration were inactive when the mean pore size was either lower than 20 microm or higher than 120 microm (Proc. Natl. Acad. Sci., USA 86(3) (1989) 933). To study the relationship between cell attachment and viability in scaffolds and the scaffold structure, CG scaffolds with a constant composition and solid volume fraction (0.005), but with four different pore sizes corresponding to four levels of specific surface area were manufactured using a lyophilization technique. MC3T3-E1 mouse clonal osteogenic cells were seeded onto the four scaffold types and maintained in culture. At the experimental end point (24 or 48 h), the remaining viable cells were counted to determine the percent cell attachment. A significant difference in viable cell attachment was observed in scaffolds with different mean pore sizes after 24 and 48 h; however, there was no significant change in cell attachment between 24 and 48 h for any group. The fraction of viable cells attached to the CG scaffold decreased with increasing mean pore size, increasing linearly (R2 = 0.95, 0.91 at 24 and 48 h, respectively) with the specific surface area of the scaffold. The strong correlation between the scaffold specific surface area and cell attachment indicates that cell attachment and viability are primarily influenced by scaffold specific surface area over this range (95.9-150.5 microm) of pore sizes for MC3T3 cells.