The linearity of mixing rates of ion-beam mixing of metals on silicon has been found to depend critically upon the method by which the mixed depth is calculated. For nonstoichiometric mixing, several methods of calculating the mixed depth may be used, namely: integrated area, halfheight, moment, error function, and 10%-90%. When applied to the same data, these methods can yield divergent results, from linear to square-root dependences of the mixed depth upon the mixing dose. For stoichiometric mixing, the calculation of mixed depth is more straightforward, but different methods of calculation still yield widely differing results. Possible causes for these discrepancies are discussed.