Some of the most polymorphic short tandem repeats (STRs) have alleles differing in size by as little as one basepair. Since sizing precision with commercially available internal standards ordinarily does not allow safe discrimination at this level, typing is accomplished through comparisons with allelic ladders run on each gel. Here we introduce the use of optimally spaced sequenced alleles as a dedicated internal standard for measurement and typing of hyperpolymorphic STRs. We have constructed such a dedicated internal standard for HUMACTBP2 ([1]; Moos and Gallwitz, EMBO I., 5 (1983) 757-761) typing, including 25 sequenced, ROX labelled HUMACTBP2 alleles spanning the size range for alleles at this locus (233-333 basepairs (bp)). Comparisons of inter-gel runs of selected alleles with this dedicated standard and the commercial GS500 internal standard demonstrate that only the presently described internal standard is suited for direct typing based on fragment length measurements only. Obvious advantages of this typing procedure are that fragment measurements are precise and in accordance with sequenced lengths, and that the typing procedure via an external allelic ladder is avoided. Sequences of the alleles used in the internal standard as well as of selected alleles for allelic ladders in two other hyperpolymorphic AAAG-repeat STRs, D11S554 (Phromchotikul et al., Hum. Mol. Genet., 3 (1991) 21) and HUMAPOA11 (Kimpton et al., PCR Methods Appl., 3 (1993) 13-22) are also presented. To allow fragment length analysis of these two STRs, five D11S554 alleles, spanning from 176 to 225 basepairs, were added to the dedicated internal standard. Performing similar comparisons as for HUMACTBP2, it is demonstrated that even if sizing accuracy is not as good as with HUMACTBP2, typing based on measured fragment size only may be achieved with both other STRs as well. Selecting different colours for the three STRs, triplex electrophoretic runs were performed of a Norwegian population database of 300 unrelated individuals. The probability that two unrelated individuals have the same type in all three STRs is 5 x 10(-7), and the combined paternity exclusion power 99.77%, indicating that this STR selection may represent a good choice for forensic genetic casework.