This study was undertaken in order to determine directly the rates of K leakage (JK) out of the tips of microelectrodes into a solution of 100 mM KCl (approximating the K concentration of the cell interior) and to relate these rates to the concentration of the filling solution and the tip resistance. The values of JK for electrodes filled with 3 M KCl having resistances of 16 and 30 M omega (when measured in 3 M KCl) were 10 and 5.5 fmol/sec, respectively. When the same electrodes were filled with 0.5 M KCl, the resistances (measured in 0.5 M KCl) increased to 62 and 115 M omega, respectively, and JK fell to 1.8 and 1.0 fmol/sec, respectively. These values are in reasonable agreement with what would be expected from theoretical considerations if leakage of KCl were the result of diffusion plus convective flow due to the hydrostatic pressure of the filling solution. We conclude that K leakage out of microelectrodes filled with 3 M KCl is unnecessarily high; leakage can be reduced fivefold by filling electrodes with 0.5 M KCl without incurring significant increases in tip or diffusion potentials or unmanageable tip resistances. Finally, the lowest rate of K leakage observed (1 fmol/sec) is still very considerable for the case of animal cells with an intracellular volume of approximately 1 pl and a K content of approximately 100 fmol. The finding of stable intracellular potentials, often for many minutes, in some tissues suggests that K which enters the cell rapidly diffuses into neighboring cells via high conductance intercellular communications.