The effect of extravascular fluid protein concentration and drug-protein binding was studied in rabbits with subcutaneously implanted Visking chambers. Chambers containing serum or saline were simultaneously studied in three animals by injecting cefazolin (30 mg/kg) intramuscularly every 4 h for four doses. Five additional animals were then studied for an 8-h period while receiving two doses of cefazolin. Peak and trough serum and chamber fluid were assayed for cefazolin concentration. The three-animal experiment demonstrated that equilibrium in the subcutaneous chambers was reached after two intramuscular injections and that the extravascular fluid levels were stable in comparison to the wide fluctuation of the serum levels. The five-animal experiment demonstrated that after two doses the serum-filled chambers contained 12.9 mug of cefazolin per ml, whereas saline-filled chambers contained 1.8 mug of cefazolin per ml. Ratios of chamber to peak serum for the second study were 12.4% for serum chambers and 1.7% for saline chambers. The cefazolin concentration in high-and low-protein extravascular fluid in both studies could be predicted from the logarithmic mean of the peak and trough serum cefazolin concentration and the antibiotic binding to rabbit serum and extravascular fluid proteins. These observations explain the previously published apparent discrepancies seen in studies of extravascular penetration of highly protein-bound antibiotics such as cefazolin.