A method of measuring net potassium flux in isolated frog sartorius muscles is described. The method depends on superfusing muscles at a slow rate in a gas phase so that a thin film of fluid is maintained on the surface. The effluent is analyzed and efflux determined by the product of flow rate and concentration change. Compartment volumes are measured and water movement is monitored by 14C-inulin. A mean resting net efflux of 25 nequiv/g/min or 0.9 pequiv/cm2/s is given in Ringers containing 2.5 mequiv/l potassium. Diffusion limits the response time of the system so that it is only possible to estimate mean flux rates over intervals of the order of 10–20 min. The minimum detectable change in potassium content of the muscle with this system is about 3.5 nequiv or approximately 0.04% of the average potassium content of the sartorius muscle. The increase in net efflux measured during activity (3.7 pequiv/cm2 per impulse) represents only about half the expected value. This is probably due to reentry of potassium with the electrochemical gradient resulting from the increased extracellular concentration. The addition of ouabain to the superfusion fluid results in an increase of potassium efflux to about four times the resting level (85 nequiv/g/min).