In measurements of grain boundary transport it is the product of the grain boundary enrichment ratio and the grain boundary diffusivity that is usually obtained. This work presents the first study in which these two terms are separated and in which the role of the grain boundary composition in grain boundary diffusion is analysed in detail. This leads to the general prediction that the grain boundary diffusion of solute and solvent will be reduced by strongly segregating solutes if they do not simultaneously enhance the bulk diffusivities. The converse occurs if the solute weakly segregates but strongly enhances the bulk diffusivities. The diffusion measurements are made in iron–tin alloys in the temperature range 563–750 °C by using radiotracers, and the segregation measurements, similarly, by Auger electron spectroscopy. The measured bulk diffusivities are similar to those found previously. The grain boundary diffusivities, determined via Suzuoka’s (1964) analysis, for iron and tin in pure iron have pre-exponential coefficients of 225 x 10
-4
and 9.2 x 10
-4
m
2
s
-1
and activation energies of 165770 and 166600 J mol
-1
respectively. Contrary to the increase in the bulk diffusivity produced by the ‘fast’ diffuser, tin, both grain boundary diffusivities are sharply reduced as the tin content rises. These and earlier results are interpreted through the effect of tin segregation on the grain boundary energy described by the theory of Borisov
et al
. (1964).