In a recent publication (D'Agostino et al., 2011) a simple equation was proposed to predict mutual diffusion coefficients in binary liquid mixtures close to the consolute point, from the tracer diffusivities and thermodynamic data. This equation is based on dynamic concentration fluctuations as predicted by critical point theory, driven by chemical potential. The equation was shown to be effective for hexane–nitrobenzene mixtures over the full range of composition and up to 15°C above the consolute temperature. In this paper, it is demonstrated that a similar equation accurately predicts mutual diffusion coefficients for the system triethylamine–water, over a wide range of concentrations and down to at least 15°C below the consolute temperature. The prediction is improved by the assumption that water moves in tightly bound dimers at all concentrations except for low water mole fractions.This is significant because, unlike hexane–nitrobenzene, triethylamine–water has a lower consolute point and the tracer diffusivities are highly non-linear as a function of concentration, showing distinct minima for both species. The agreement of prediction with experiment for these two systems supports the idea that dynamic concentration fluctuations are driven by chemical potential gradients; this being the basis of the equations used to predict mutual diffusion coefficients in the mixtures. It raises the question of how generally this may be applicable to non-ideal mixtures not close to their consolute point, even mixtures which do not exhibit liquid–liquid phase separation, a topic which will be addressed in future publications. The possibility of predicting mutual diffusion coefficients from tracer diffusivities has practical implications, particularly where measurement of the mutual diffusion coefficient is difficult; tracer diffusivities can be easily measured by pulsed field gradient NMR or estimated using molecular simulation calculations.