Low temperature ion beam mixing rates for Cu‐Ti, Ni‐Ti, and Fe‐Ti layers have been found to be significantly lower than predicted by a popular semi‐empirical thermal spike model. It has been proposed that the unavoidable hydrogen contamination of the as‐deposited Ti films may have reduced the mixing rates, but the measurement of even lower mixing rates for Fe‐V and Fe‐Co bilayers shows the discrepancy to be more fundamental. Still, a systematic dependence on heat of mixing suggests that some sort of diffusional (thermal spike?) mechanism is involved.