The ratio of sodium to potassium in the exosphere of Mercury is highly variable and averages about 100, considerably larger than for the Moon. We considered the possibility that this high value results from potassium loss rates that are larger than sodium loss rates. The principal loss processes are the sweeping of hot atoms off the planet by solar radiation acceleration and the capture of metal photoions by the solar wind. We found that radiation acceleration losses are not sufficiently different for sodium and potassium to explain the high ratio. Capture of sodium and potassium photoions by the solar wind is thought to be the major loss process. However, there is a competing process for the ions, which is collision with the surface, followed by neutralization. These atoms can return to the exosphere. The net steady state density of sodium and potassium in the exosphere depends on the relative efficiencies of solar wind capture and surface recycling. We propose that the photoion loss and recycling rates are sufficiently different for sodium and potassium to account for the high ratio. The higher mass of the potassium atom relative to sodium results in a smaller scale height and a larger gyroradius for potassium relative to sodium, which may result in more rapid net loss of potassium.