Fluoromisonidazole labeled with H-3 or F-18 has been tested as a quantitative probe for hypoxic cells in vitro and in rodent and spontaneous dog tumors in vivo. In V-79, EMT-6(UW), RIF-1, and canine osteosarcoma cells in vitro, the binding of 50 microM [H-3]Fluoromisonidazole was 50% inhibited by 1000-2000 ppm O2, relative to binding under anoxic conditions. After a 3 hr incubation with labeled drug, the anoxic/oxic binding ratios ranged from 12 to 27 for the four cell types. Retention of [H-3]fluoromisonidazole 4 hr after injection was greater in large KHT tumors (400-600 mm3) with an estimated hypoxic fraction greater than 30%, than in smaller tumors (50-200 mm3) with an estimated hypoxic fraction of 7-12%. RIF-1 tumors, with an estimated hypoxic fraction of 1.5%, retained the least label, with tumor: blood ratios ranging from 1.7 to 1.9. Spontaneous dog osteosarcomas were imaged with a time of flight positron emission tomograph for up to 5 hr following injection of [F-18] fluoromisonidazole. Analysis of regions of interest in images allowed creation of dynamic tissue time activity curves and calculation of tissue uptake in cpm/gram. These values were compared to radioactivity in plasma. In all cases, retention in some tumor regions exceeded that in plasma and in normal tissue, such as muscle or brain, by 3 to 5 hr post injection. Uptake of fluoromisonidazole in tumors was heterogeneous, with ratios of maximum to minimum uptake as high as 4 in different regions of interest in the same tumor. Tumor:plasma values ranged from 0.28 to 2.02. The oxygen dependency of fluoromisonidazole retention was similar in a variety of cell types and was 50% inhibited by O2 levels in the transition between full radiobiological hypoxia and partial sensitization. The quantitative regional imaging of [F-18] fluoromisonidazole in spontaneous canine tumors at varying times post-injection lays the basis for imaging and modeling of oxygen-dependent drug retention in different regions of human neoplasms.