In Doressoundiram et al. (Icarus, 2010), we presented and discussed two sets of observations of the Na/K ratio in Mercury's exosphere. These observations suggested that the different spatial distributions of Mercury's Na and K exospheres could help explain the anomalously large values of Na/K ratio observed at Mercury. In Leblanc and Johnson (Icarus 2010), we simulated the annual cycle of Mercury's sodium exosphere, emphasizing the importance of the day to night side migration of the sodium and the close relation between the content of the sodium exosphere and the content of the surface reservoir of adsorbed sodium. In this paper, the potassium exosphere is modelled using laboratory measurements to constrain the various ejection processes. The annual cycle of the exospheric potassium emission brightness is compared to the few available observations showing that our model provides a reasonable description of this exospheric component. A comparison with the best spatially resolved observation of Doressoundiram et al. (Icarus, 2010) highlights key features of the Na/K spatial distributions. It is essentially differences in global transport, the loss rate and the desorption efficiencies which account for the spatial distribution of the observed ratio. The Na/K ratio in the exosphere cannot be simply associated with the primary abundances but is closely related to the relative loss rates. Accounting for the transport and loss, the observed Na/K exospheric ratio may be consistent with an initial abundance close to solar or meteoritic ones.