Abstract In this paper the correlation factors, as obtained from the ratio of tracer diffusion coefficients in the binary random alloy model, are compared with the correlation factors as would be obtained from isotope effect measurements. In this context, two theories, one due to Manning, the other due to Moleko, Allnatt and Allnatt are examined in detail. We show that for both atomic components Manning's theory (where both types of correlation factors are automatically equal) gives reasonable agreement with Monte Carlo data only when the tracer diffusion coefficients are within a factor of two. On the other hand, the theory of Moleko et al. gives excellent agreement for both correlation factors over a very wide range of diffusion coefficient ratios. To demonstrate some of these differences the formalisms of Manning and Moleko et al. are applied to the Fe—Co and Cu—Zn systems.