In mutation and cell transformation assays, it has long been recognized that the common practice of using different numbers of cells on dishes with or without selective conditions creates a source of bias in mutant fraction determination. This is simply because colony formation may be enhanced or suppressed at higher initial cell densities, depending on the assay and agent tested. We propose a solution that consists of the inclusion of an experimentally distinguishable population of cells as an internal standard for colony-forming ability at the high cell density required for detection of rare variants. This method is found to be highly satisfactory for use in measuring mutation to 6-thioguanine resistance in a diploid human B lymphoblast line. For treatment with anti-2,3-dihydroxy-1,10b-epoxy-1,2,3-trihydrofluoranthene (FDE), N-methyl-N'-nitro-N-nitrosoguanidine (MNNG), and 4-nitroquinoline-oxide (4NQO), the calculated induced mutant fractions using the internal-standard method were significantly lower than those calculated using the conventional low-density-plating efficiency method. The results of these experiments and our analysis lead us to conclude that this approach is applicable to all single cell mutation or transformation assays and is a necessary feature of assays in which an accurate knowledge of the fraction of rare variants is required.