Paternity testing using DNA polymorphism of variable numbers of tandem repeat (VNTR) regions with restriction fragment length polymorphism (RFLP) was implemented. HinfI-digested DNA was separated by electrophoresis in agarose gels and hybridized with radiolabelled probes detecting the VNTR-systems D2S44 (YNH24), D5S43 (MS8), D7S21 (MS31), D7S22 (g3), and D12S11 (MS43a). The intra gel variability of 970 duplicate investigations on the same gel of DNA from 122 individuals showed no differences exceeding 1.25 mm between the positions of the corresponding DNA fragments. The comparison of 1,624 DNA fragments from 342 mother/child pairs showed only one difference above 1.25 mm which was interpreted as a mutation. Based on these observations, we decided to consider an intra gel difference above 1.25 mm between the non-maternal DNA fragment of the child and the nearest DNA fragment of the putative father as an exclusion in paternity testing. This matching criterion was used for the comparisons of 1,197 DNA fragment differences in 247 pairs of children and putative fathers who had not been excluded by conventional marker systems. In all of these cases, the migration differences between the DNA fragments of non-excluded men and the DNA fragments of the children were less than 1.25 mm except in 6 cases (0.5%). The man/child differences in all of 227 false trios exceeded 1.25 mm in 2 or more of the 5 VNTR systems investigated. Matching criteria for inter gel comparisons in paternity testing were established. The frequency distribution of HinfI digested DNA fragments of the 5 VNTR systems in 650 unrelated Danes is presented and the raw data is available.