In this paper, an extended Maxwell-Hellawell numerical grain size prediction model is employed to predict as-cast grain size of inoculated aluminum alloys. Given melt composition, inoculation and cooling conditions, the model is able to predict maximum nucleation undercooling, cooling curve and final as-cast grain size of multi-component alloys. The proposed model has been applied to various binary and multicomponent alloys. Upon analyzing the numerical simulation results, it is found that for both binary and multi-component alloys, grain size does not have a one-to-one relation with Growth Restriction Factor, Q, but has a clear ubiquitous correlation with the average diffusivity-weighted Q, defined as R in this paper. This founding helps solve the controversy seen in the recent work on analytical grain size and Q relations. It also has been used to interpret the scatters seen in the measured grain size as a function of and Q valuesdata reported in the literature.