Techniques were developed to determine the location of exchangeable K pools that had been identified previously in kinetic studies of the intact perfused bullfrog kidney. Following perfusion of the kidneys with 42K, a washout of the isotope was begun and interrupted at various times; the kidneys were removed, frozen, dried at low pressure and temperature, and then microdissected. Glomerular capillary tufts, small segments of tissue containing early distal tubules (diluting segment), and other segments containing proximal tubular convolutions were removed and analyzed for total content of K and 42K. In the intact kidney 77% of tissue K exchanged in 60 min. The exchangeable K concentration was 95 mu eq/ml cell water. Correction for the K activity coefficient in Ringer solution yielded an activity of 72 mu eq/ml. Thirty-two percent of glomerular capillary K exchanged in 60 min; 7% exchanged with a half time of 3.5 min; and the remainder exchanged at a rate too slow to measure. The data from tissue containing proximal tubular segments were too scattered to permit analysis. In segments containing early distal tubules, 67% of tissue K was contained in two exchangeable pools: one pool exchanged at a rate 10-fold greater than did the other. The data for these two distal pools were analyzed in terms of a parallel model (two cell types?) and a nested model (cytoplasm and subcellular organelles?). Pool size and exchange rates were calculated for both models. Electron microscopic analysis revealed that early distal tubular segments contain only one cell type which has a large population of mitochondria. This suggests that the nested model is more plausible. The fast distal pool exchanged at the same rate as the fast-exchanging pool identified in kinetic studies of the intact functioning kidney and is considered to be the K secretory pool.